Design of a Hand-Launched Solar-Powered Unmanned Aerial Vehicle (UAV) System for Plateau

applied sciences

Article Design of a Hand-Launched Solar-Powered Unmanned Aerial Vehicle (UAV) System for Plateau

Xin Zhao 1, Zhou Zhou 1,*, Xiaoping Zhu 2 and An Guo 1

1 School of Aeronautics, Northwestern Polytechnical University, Xi’an 710072, China; [email protected] (X.Z.); [email protected] (A.G.) 2 Science and Technology on UAV Laboratory, Northwestern Polytechnical University, Xi’an 710072, China; [email protected] * Correspondence: [email protected]

Received: 8 January 2020; Accepted: 7 February 2020; Published: 14 February 2020

Abstract: This paper describes our work on a small, hand-launched, solar-powered unmanned aerial vehicle (UAV) suitable for low temperatures and high altitudes, which has the perpetual flight potential for conservation missions for rare animals in the plateau area in winter. Firstly, the conceptual design method of a small, solar-powered UAV based on energy balance is proposed, which is suitable for flight in high-altitude and low-temperature area. The solar irradiance model, which can reflect the geographical location and time, was used. Based on the low-temperature discharge test of the battery, a battery weight model considering the influence of low temperature on the battery performance was proposed. Secondly, this paper introduces the detailed design of solar UAV for plateau area, including layout design, structure design, load, and avionics. To increase the proportion of solar cells covered, the ailerons were removed and a rudder was used to control both roll and yaw. Then, the dynamics model of an aileron-free layout UAV was developed, and the differences in maneuverability and stability of aileron-free UAV in plateau and plain areas were analyzed. The control law and trajectory tracking control law were designed for the aileron-free UAV. Finally, the flight test was conducted in Qiangtang, Tibet, at an altitude of 4500 m, China’s first solar-powered UAV to take off and land above 4500 m on the plateau in winter ( 30 C). The test data − ◦ showed the success of the scheme, validated the conceptual design method and the success of the control system for aileron-free UAV, and analyzed the feasibility of perpetual flight carrying different loads according to the flight energy consumption data.

Keywords: solar-powered UAV; perpetual flight; conceptual design; aileron-free; flight control system; flight test

1. Introduction With the development of the world economy, energy crisis and environmental pollution are becoming a serious problem. In recent years, with the rapid improvement of technology of solar cell and secondary battery, solar-powered aircraft, due to the advantages of environmental protection and long endurance, following the ‘Pathfinder’ [1] and ‘Helios’ [2], the world set off a research upsurge of solar-powered UAV once again. Solar-powered UAV, which converts solar energy into electricity and stores it in secondary batteries during daytime flight, has better endurance than conventional aircraft and may even be able to fly permanently. The long-endurance capability of the solar-powered UAV platform is particularly important in rescue missions, forest fire prevention, communication coverage, and protection of endangered animals. In view of the extreme flight environment in the plateau region, this paper proposes a design method of small, solar-powered UAV suitable for low-temperature and high altitude areas, including a solar irradiance model that can reflect the geography and time and a

Appl. Sci. 2020, 10, 1300; doi:10.3390/app10041300 www.mdpi.com/journal/applsci Appl.Appl. Sci. Sci.2020 2020, 10, ,10 1300, x FOR PEER REVIEW 2 of2 28 of 26

can reflect the geography and time and a battery weight model that considers the impact of low batterytemperature weight on model battery that performance. considers the Based impact on ofthis, low a temperaturesmall aileron-free on battery UAV was performance. developed Basedfor onecological this, a small monitoring aileron-free and UAVanimal was protection developed in forplateau ecological areas. monitoringThis paper andalso animal developed protection the indynamics plateau areas. model This of paper aileron-free also developed UAV, analyzed the dynamics the characteristics model of aileron-free of UAV’s UAV, stability analyzed and the characteristicsmaneuverability of UAV’s in high-altitude stability and maneuverabilityflight, and designed in high-altitude the control flight, and and tracking designed system the control for andhigh-altitude tracking system flight. for Finally, high-altitude in Qiangtang, flight. Finally,Tibet, China’s in Qiangtang, first solar-powered Tibet, China’s UAV first took solar-powered off and UAVlanded took at off anand altitude landed ofat more an altitude than 4500 of morem. The than experiment 4500 m. Theverified experiment the correctness verified of the the correctness design of themethod design and method the control and themethod, control and method, analyzed and the analyzed feasibility the of feasibility the perpetual of the flight perpetual with different flight with loads according to the flight energy consumption data (Figure 1). different loads according to the flight energy consumption data (Figure1).

(a) (b) (c)

FigureFigure 1. The1. The solar-powered solar-powered unmanned unmanned aerial aerial vehicle vehicle (UAV) (UAV) used used forecological for ecological monitoring monitoring and animaland protection:animal protection: (a) Mini-Phantom (a) Mini-Phantom (MP) solar-powered (MP) solar-powered UAV, (b) infrared UAV, ( podb) infrared for ecological pod for monitoring, ecological (c) Tibetanmonitoring, antelopes (c) Tibetan photographed antelopes by photogra optical camera.phed by optical camera.

InIn recent recent years, years, solar-powered solar-powered aircraftaircraft havehave been developed in in both both manned manned and and unmanned unmanned aircraft.aircraft. In termsIn terms of mannedof manned solar solar aircraft, aircraft, ‘Solar ‘Sol Impulsear Impulse 2’, the 2’, world’s the world’s largest largest solar-powered solar-powered airplane, successfullyairplane, successfully completed completed the world’s the first world’s solar-powered first solar-powered flight around flight around the world the onworld 26 Julyon 26 2016 July [ 3]. On2016 the other[3]. On hand, the manyother hand, researchers many focusresearchers on employing focus on large-scaleemploying solar-poweredlarge-scale solar-powered high-altitude long-endurancehigh-altitude long-endurance (HALE) UAVs as(HALE) atmospheric UAVs as satellites. atmospheric Typical satellites. examples Typical are ‘Solara’examples [4 ]are and ‘Solara’ ‘Zephyr S’,[4] the and latter ‘Zephyr of which S’, the has latter broken of wh theich previous has broken record the of previous 14 days record set by theof 14 ‘Zephyr days set 7’ by prototype the ‘Zephyr [5] with7’ prototype [5] with approximately 25 days and 23 h of flight on 11 July 2018 [6]. In contrast, approximately 25 days and 23 h of flight on 11 July 2018 [6]. In contrast, smaller-scale, solar-powered smaller-scale, solar-powered UAVs are mostly designed for low-altitude long-endurance (LALE) UAVs are mostly designed for low-altitude long-endurance (LALE) applications. In 2008, ‘sky sailor’, applications. In 2008, ‘sky sailor’, a small solar-powered UAV from Switzerland, flew for 27 h. The a small solar-powered UAV from Switzerland, flew for 27 h. The structural design of the UAV referred structural design of the UAV referred to the glider for competition and it weighed only 780 g [7]. The to the glider for competition and it weighed only 780 g [7]. The ‘AtlantikSolar’ small UAV from ‘AtlantikSolar’ small UAV from the Swiss Federal Institute of Technology Zurich completed a flight theof Swiss 81.5 h Federal in July 2015, Institute setting of Technologyan endurance Zurich record completed for an aircraft a flight under of 81.550 kg. h Compared in July 2015, with setting large an endurancesolar-powered record aircraft, for an aircraftsmall solar-powered under 50 kg. UAV Compared have higher with largestructural solar-powered weight coefficient aircraft, and small solar-poweredlower load UAVcapacity, have so higher theystructural are suited weight for low-altitude coefficient and and lower long-endurance load capacity, missions. so they are Noth suited forproposed low-altitude a matching and long-endurance parameter mo missions.del of energy Noth and proposed power afor matching perpetual parameter flight [7] modelof small of solar energy andUAV power in 2008. for perpetualBased on the flight energy [7] of system small model, solar UAV Philipp in 2008.Oettershagen Based onproposed the energy an energy-robust system model, Philippdesign Oettershagen method that proposedallows design an energy-robust of solar-powe designred UAVs method for energetically that allows robust design perpetual of solar-powered flight UAVsin sub-optimal for energetically meteorological robust perpetual conditions flight [8]. inIn sub-optimal both methods, meteorological the solar irradiance conditions model [8]. was In both a methods,simplified the solarsinusoidal irradiance model. model Small was solar-powere a simplifiedd sinusoidalUAVs fly model.mostly Small at low solar-powered altitudes, so UAVs most fly mostlyprevious at low studies altitudes, did not so consider most previous the effect studies of low did temperatures not consider on battery the effect performance. of low temperatures on batterySection performance. 2 presents the conceptual design method of perpetual flight suitable for flight on low temperatureSection2 presents and high the altitude conceptual based designon energy method balance, of perpetual and explores flight the suitable feasibility for of flight perpetual on low temperatureflight of small and highhand-launched altitude based solar-powered on energy balance, UAV in and plateaus explores such the feasibilityas Tibet, China, of perpetual which flight is ofparticularly small hand-launched important for solar-powered the protection UAV of rare in plateausanimals such such as as the Tibet, Tibetan China, antelope. which In is this particularly paper, importantirradiance for predicted the protection model of[9] rare was animals adopted such to establish as the Tibetanthe solar antelope. energy model, In this which paper, can irradiance more accurately reflect the solar irradiance at different geographical locations and time. The accuracy of predicted model [9] was adopted to establish the solar energy model, which can more accurately reflect the irradiance model was verified by experimental data, which were measured by the total solar the solar irradiance at different geographical locations and time. The accuracy of the irradiance model was verified by experimental data, which were measured by the total solar irradiance meter. In this Appl. Sci. 2020, 10, 1300 3 of 26 part, the low-temperature discharge test was carried out to test the influence of low temperature on the battery discharge performance, and the battery weight model considering the influence of low temperature was proposed. Section3 presents the detailed design of the Mini-Phantom (MP) solar UAV. The MP UAV system design includes layout design, structure design, payloads, and avionics selection. Sections4 and5 presents the dynamic model and control algorithms, in which the model of the high-altitude propulsion system was modified through the low-altitude dynamic experiment. Based on the motion equation of UAV, this section analyzes the change of flight quality with altitude and velocity when flying in plateaus and plain areas. In addition, longitudinal height and lateral trajectory tracking controllers were designed based on successive loop closure for plateau flight environment. Flight tests were conducted on the Tibetan plateau. Experimental results verified the correctness of the design method and control algorithm, and analyzed the feasibility of permanent flight in the plateau. Section6 contributes an outlook into the specific condition and the whole process is summarized.

2. Conceptual Design Method To achieve perpetual flight, the solar-powered UAV was designed based on weight balance and energy balance. The former determined whether the plane would fly as expected. In the latter, the balance between the energy collected by the solar cells and the energy consumed in the flight determined the endurance of the UAV. In order to simplify the model, UAV’s cruise state was steady level flight at the same altitude. During the cruise, there was no energy conversion through climbing and descending. In design, the first consideration was the balance of forces. Similar to conventional long-endurance aircraft, solar-powered UAV have a single flight state, which is a ‘single design point’ flight platform. The balance between lift and weight and the balance between thrust and drag are necessary during steady flight. Power balance is also important in the design of solar-powered aircraft. In order to achieve the goal of perpetual flight, the total energy collected by the solar cells during the day flight must be more than the energy consumed during the flight. Besides, the remaining energy from the batteries should be sufficient for the night flight at the end of the day. In this paper, the analytical and continuous approximation method was used, that is, the model describing the characteristics of each part was used to establish the relationship between each part model by analytic equation. The advantage of this method is that the analytical solution can be obtained directly and the unique optimal design can be obtained directly. However, the method needs accurate mathematical models. In this paper, first of all, the power expression of the plane in cruise was established, and then the total energy collected by the solar cells during the day flight was obtained based on the solar irradiance model. Then, the weight prediction model of each part of the aircraft was given, and a closed loop was formed according to the relationship between the energy, so as to obtain the analytical solution that makes the scheme feasible.

2.1. Power of Level Flight

The lift and drag of the aircraft can be obtained from Formula (1), where CL and represent CD the lift and drag coeﬃcients, respectively. ρ is air density and S is wing area. According to these two basic formulas, the mechanical power consumed during cruise can be obtained, as shown in Formula (2).

( ρ 2 W = L = 2 V SCL ρ 2 (1) T = D = 2 V SCD s mg C 2ARg3 m3/2 P = TV = V = D (2) lev (L/D) 3/2 ρ b CL Appl. Sci. 2020, 10, 1300 4 of 26

The total electrical power of the aircraft during cruise Pelec_tot included mechanical power and electrical power. The efficiency of the motors ηmot, electronic controllers ηctrl, and propellers ηplr, as well as the power consumption of avionics Pav and payload instrument Ppld must be considered. Since the battery voltage was higher than the voltage of avionics and payload instrument, the efficiency of step-down ηbec, also called battery elimination circuit (BEC), should also be considered. In order to improve the reliability and maintainability, MP UAV’s propulsion system was propeller-motor direct drive. The efficiency of gearbox can be ignored. The total electrical power during cruise is shown in Formula (3). 1 1 Pelec_tot = Plev + Pav + Ppld (3) ηctrlηmotηplr ηbec The total energy consumed during the all-day flight is shown in the Formula (4). Unlike daytime flight, all energy consumed during the night flight was from the battery, so the charging efficiency ηchrg and ηdchrg discharging efficiency of battery also need to be considered. ! Tnight Eelec_tot = Pele_tot Tday + (4) ηchrgηdchrgktemp where Tday and Tnight present day time and night time respectively, and ktemp presents the temperature influence factor. At dusk and dawn, when the solar energy collected by solar cells is less than the required power, the energy comes from both the solar cells and battery, and the conversion from one source to another is gradual. The process of energy source conversion is seen as instantaneous to simplify the calculation.

2.2. Solar Energy Obtained during the Day In this paper, solar irradiance predicted model was used to estimate the solar irradiance. The solar energy obtained throughout the day can be calculated by solar irradiance model. Irradiance depends on many variables, such as geographical location, time, plane orientation, weather conditions, and albedo, which represents the reﬂection of ground [10]. In general, the total irradiance Itot consists of direct irradiance Ib and scattering irradiance Id. The direct irradiance [10] is shown in Formula (5).         h    cs exp    − hs   I = I0 exp h    " + !#ss+   − αs αdep hb   sin  (5)  1+αdep/90   = [ + ( )]  I0 Gsc 1 0.033 cos 360nd/365   αdep = 0.57 + arccos(Reth/(Reth + h))

2 where Gsc is the standard solar radiation constant, taking 1367 W/m . I0 is the solar irradiance outside the atmosphere on the nd day of the year, cs and ss are both constants, taking 0.357 and 0.678, respectively. The h is the ﬂight altitude of solar aircraft and αdep is the modiﬁed value relative to ground. Reth is the radius of the Earth. The αs is determined by local latitude ϕlat, solar declination angle δs, and solar hour angle θh. The relationship is shown in Formula (6).   sin αs = sin ϕlat sin δs + cos ϕlat cos δs cos θh  (6)  nd+284  δs = 23.45 sin 360 365 in which the corresponding relationship between solar hour angle θh and solar hour Hs is shown in Formula (7). θ = 15(Hs 12) (7) h − Appl. Sci. 2020, 10, 1300 5 of 26

Solar time in any region can be calculated by Formula (8).   = + Lst Lct + Et  Hs Hct 15− 60   Et = 0.0172 + 4.28 cos B 7.35 sin B 3.35 cos 2B 9.732 sin 2B (8)  − − −  B = 360 (n 1) 365 d − where Hct is the local time. Lst is the standard longitude adopted by Hct. Lct is the local longitude. Et is the correction between Hs and Hct caused by the revolution of earth. It is assumed that the scattering irradiance is omnidirectional and linearly proportional to the direct irradiance. Here, the proportionality coeﬃcient is 0.08 [9]. The relationship between scattering irradiance and direct irradiance is shown in Formula (9). ! h Id = 0.08Ib exp (9) −hs The total solar irradiance considering latitude, altitude, and ﬂight time is expressed as Formula (10).

Itot = Ib sin αs + Id (10)

The solar irradiance of Xi’an, China, on 27 May 2017 at an altitude of 400 m, simulated by the irradiance prediction model, is shown in the Figure2a. The measured data in the ﬁgure was measured on 27 May 2017. For the parameters, such as the maximum irradiance, the corresponding time, and the timeAppl. Sci. of the2020, sun 10, x setting, FOR PEER the REVIEW irradiance model ﬁt well with the measured data. Figure2b presents6 of the 28 measuring instruments of solar irradiance.

(a) (b)

Figure 2.2. ((aa)) ComparisonComparison of of predicted predicted and and measured measured values values of solar of solar irradiance irradiance on 27 on May 27 2017,May Xi’an,2017,

ChinaXi’an, China (34◦16’ (34°16’ N 108 ◦N54’ 108°54’ E). (b) E). Measuring (b) Measuring instrument instrument of solar of irradiance. solar irradiance.

Obviously, energyenergy density density captured captured by by solar solar cells cells is theis the integral integral of irradiance of irradiance with with respect respect to time, to intime, which in which energy energy density density is the ratiois the of ratio captured of captur energyed energy to solar to cell solar area. cell Considering area. Considering the actual the layingactual area of solar cells Asc, packaging efficiency η , electricalη energy conversion efficiency ηsc, maximumη laying area of solar cells Asc , packaging efficiencycbr cbr , electrical energy conversion efficiency sc , power point tracker (MPPT) efficiency η , and weather influence factor, the solar energy captured maximum power point tracker (MPPT)mppt efficiencyη , and weather influence factor, the solar by solar cells of the aircraft is shown in Formula (11). mppt energy captured by solar cells of the aircraft is shown in Formula (11). Z EsunEItdtA_tot = = Itot(()t)dtAscηηηηηwthrηscηcbrηmppt (11) sun_ tot tot sc wthr sc cbr mppt (11)

2.3. Weight Estimation Model 2.3. Weight Estimation Model The total takeoff weight of an aircraft consists of the weight of components, which directly affects The total takeoff weight of an aircraft consists of the weight of components, which directly the flight power. In order to calculate the flight power accurately, the weight estimation model should affects the flight power. In order to calculate the flight power accurately, the weight estimation be precise as much as possible. The weight of the main components of the UAV includes weight of model should be precise as much as possible. The weight of the main components of the UAV includes weight of mission payloads mpld and avionics mav , structural weight maf , weight of solar cells msc , weight of MPPT mmppt , weight of batteries mbat , and weight of propulsion system mprop . Payload and avionics are fixed weights, which are related to mission requirements and shown in Formula (12). The weight of other components is related to the parameters of UAV size, flight power, etc. =+ mmmfixed av pld (12)

The weight of aircraft structures is difficult to model accurately. Most structural weight estimation models are empirical formulas, which are based on a large amount of data from existing aircraft. Romeo [11], Youngblood [12], and Rizzo [13] proposed several weight estimation models for large wingspan, solar-powered aircraft, which used wing area and wingspan as variables to estimate the structural weight of the aircraft. Noth [7] screened the data by wingload and structure weight to modify the parameters of the weight estimation model, which can apply to the small, light-structure, solar-powered UAV (Formula (13)).

xx = 21 mkARbaf af (13)

The power captured by the solar cells is linear with the laid area. The area of the solar cells determines whether the energy captured from sun is enough to balance the energy consumed in flight. The area and weight of solar cells can be calculated based on the energy–power balance, which is shown in Formula (14). Appl. Sci. 2020, 10, 1300 6 of 26

mission payloads mpld and avionics mav, structural weight ma f , weight of solar cells msc, weight of MPPT mmppt, weight of batteries mbat, and weight of propulsion system mprop. Payload and avionics are ﬁxed weights, which are related to mission requirements and shown in Formula (12). The weight of other components is related to the parameters of UAV size, ﬂight power, etc.

m f ixed = mav + mpld (12)

The weight of aircraft structures is diﬃcult to model accurately. Most structural weight estimation models are empirical formulas, which are based on a large amount of data from existing aircraft. Romeo [11], Youngblood [12], and Rizzo [13] proposed several weight estimation models for large wingspan, solar-powered aircraft, which used wing area and wingspan as variables to estimate the structural weight of the aircraft. Noth [7] screened the data by wingload and structure weight to modify the parameters of the weight estimation model, which can apply to the small, light-structure, solar-powered UAV (Formula (13)). x2 x1 ma f = ka f AR b (13) The power captured by the solar cells is linear with the laid area. The area of the solar cells determines whether the energy captured from sun is enough to balance the energy consumed in ﬂight. The area and weight of solar cells can be calculated based on the energy–power balance, which is shown in Formula (14).  Tnight  = 1 R +  Asc Tday η η k Pelec_tot ηscηcbrηmpptηwthr Itot charg dcharg temp (14)   msc = Asc(ksc + kenc) where ksc presents the surface density of solar cells. However, exposed solar cells, which are fragile and oxidizing, need to be encapsulated for practical use. Surface density of packaging kenc also needs to be considered. MPPT is used to track the maximum power by adjusting the voltage of the solar cells, and its weight is proportional to the maximum power, that is, proportional to the area of the solar cells. The weight of MPPT is shown in Formula (15), where Psolmax presents the maximum input power of solar cells, obtained by measurement.

mmppt = kmpptPsolmax = kmpptImaxηscηcbrηmpptAsc (15)

For batteries, their weight is proportional to the energy stored and inversely proportional to energy density kbat, which is an important parameter to evaluate battery performance. However, the energy density of the battery is inversely proportional to the discharge power. For solar-powered aircraft that focus on cruise, the constraint of discharge power can be ignored. Therefore, high-energy density batteries, such as lithium-ion batteries, should be selected. For perpetual flight, the requirement of minimum battery weight is that the energy stored by batteries is sufficient for night flight, which is shown in Formula (16). Tnight mbat = Pelec_tot (16) ηdchrgkbat Another factor that cannot be ignored that affects battery performance is ambient temperature. For the lithium-ion batteries concerned in this paper, generally speaking, the most suitable operating temperature is 0–40 ◦C. When the temperature is lower than 0 ◦C or higher than 40 ◦C, the battery discharge efficiency will decrease sharply, which means the reduction of battery capacity or the increase of battery weight required for aircraft parameters. The flight environment of MP solar UAV was the plateau area in winter. In order to study the influence of low temperature on battery capacity, the low-temperature discharge test of energy system Appl. Sci. 2020, 10, x FOR PEER REVIEW 7 of 28

  1 Tnight  =+ ATPsc day elec_ tot ηη η η I ηη k  sc cbr mppt wthr tot charg dch arg temp (14)  =+() mAkksc sc sc enc where ksc presents the surface density of solar cells. However, exposed solar cells, which are fragile and oxidizing, need to be encapsulated for practical use. Surface density of packaging kenc also needs to be considered. MPPT is used to track the maximum power by adjusting the voltage of the solar cells, and its weight is proportional to the maximum power, that is, proportional to the area of the solar cells. The weight of MPPT is shown in Formula (15), where Psol max presents the maximum input power of solar cells, obtained by measurement. ==ηη η mkPkImppt mppt solmax mppt max sc cbr mppt A sc (15)

T mP= night batη elec_ tot (16) dchrgk bat

Another factor that cannot be ignored that affects battery performance is ambient temperature. For the lithium-ion batteries concerned in this paper, generally speaking, the most suitable operating temperature is 0–40 °C. When the temperature is lower than 0 °C or higher than 40 °C, the battery discharge efficiency will decrease sharply, which means the reduction of battery capacity or the increase of battery weight required for aircraft parameters. Appl. Sci. 2020, 10, 1300 7 of 26 The flight environment of MP solar UAV was the plateau area in winter. In order to study the influence of low temperature on battery capacity, the low-temperature discharge test of energy wassystem conducted was conducted in a low-temperature in a low-temperature and low-pressure and testlow-pressure chamber, experimentaltest chamber, instruments experimental are instrumentsshown in Figure are shown3. in Figure 3.

Figure 3. The discharge test of energy system in low-temperature and low-pressure test chamber.

The test subjects included propulsion systems (brushless motor, propeller, and motor controller), autopilot systems, and a lithium-ion battery rated at 230 Wh for low-temperature resistance. The test chamber provided a 25 C, 540 hPa low-temperature, and low-pressure simulation environment. − ◦ During the test, the propulsion system operates with a constant throttle command until the voltage reached the battery’s protective voltage, and the sensor recorded the output power, voltage, and current of the battery bus. Figure4 shows the test results of the low-temperature discharge test, which are the changes of power, voltage, and current with time, respectively. In the test, the system worked normally under low-temperature conditions, the voltage fluctuation range was small, basically linear decline. The amplitude of current fluctuation was about 0.4 A and the smoothed current-time curve also presented a linear decline. By integrating the power-time curve discretely, the total output energy of the battery in the test can be obtained, which was about 159.01 Wh. At a low temperature of 25 C, the discharge − ◦ efficiency of the battery was about 69.1%. In the discharge experiment at room temperature (20 ◦C), the output energy was about 220.73 Wh and the discharge efficiency of the battery was about 95.9%. Therefore, solar-powered UAV flying in extreme environments, temperature impact factor ktemp must be considered in the design process. In this test, the temperature influence factor was about 0.72. In order to prevent the low temperature from damaging the battery performance, the weight of the insulation material or the temperature control system of the battery also needs to be considered in the high-altitude, solar-powered UAV design. The battery weight model considering the effect of low temperature is shown in the Formula (17).

Tnight mbat = Pelec_tot (17) ηdchrgkbatktemp

Electric propulsion systems usually contain propellers, motors, controllers, gearbox, etc. For small, solar-powered UAVs, a gearbox is not necessary to reduce the system complexity and cost. In order to simplify the model, the propulsion system is considered as a whole, and the weight of the propulsion system is positively correlated with the maximum power. Factor kprop is used to predict the Appl. Sci. 2020, 10, 1300 8 of 26 weight of the propulsion system, whose value is obtained by statistical data. The weight model of Appl. Sci. 2020, 10, x FOR PEER REVIEW 8 of 27 propulsion system is shown in Formula (18). The test subjects included propulsion systems (sbrushless motor, propeller, and motor 3 3/2 controller), autopilot systems, and a lithium-ion batteryCD 2 ratedARg atm 230 Wh for low-temperature controller), autopilot systemsm,prop and= akprop lithiumP =-ionkprop battery rated at 230 Wh for low-temperature(18) lev 3/2 ρ b resistance. The test chamber provided a −25 °C, C 540L hPa low-temperature, and low-pressure simulation environment. During the test, the propulsion system operates with a constant throttle commandThe total until energy the voltage consumed reach duringed the day battery and night’s protective ﬂight can voltage, be obtained and by the the sensor aerodynamic recorded model the outputand the power, weight voltagemodel system., and current The closed of the loop battery of the bus.energy Figure balance 4 show is reconstructeds the test results by the weight of the lowmodel-temperature of each component. discharge test, Figure which5 presents are the a detailedchanges loopof power, that visually voltage, represents and current the with conceptual time， respectively.design process of the solar-powered UAV in this paper.

(a)

(b)

(c)

FigureFigure 4. 4. TestTest result result of of low low-temperature-temperature discharge discharge test test at at−25 25degrees degrees Celsius, Celsius, 540 540hPa hPa:: (a) Power (a) Power,, (b) − v(boltage) voltage,, (c) c (urrent.c) current.

In the test, the system worked normally under low-temperature conditions, the voltage fluctuation range was small, basically linear decline. The amplitude of current fluctuation was about 0.4 A and the smoothed current-time curve also presented a linear decline. Appl. Sci. 2020, 10, 1300 9 of 26 Appl. Sci. 2020, 10, x FOR PEER REVIEW 10 of 28

FigureFigure 5. 5. ConceptualConceptual design design progress progress of of solar-powered solar-powered UAV UAV based on energy balance.

TheThe parameters contained in Figure 55 cancan bebe divideddivided intointo threethree categories.categories. TheThe first ﬁrst part is the weight prediction model, which consists of of the weight models of each component.component. The The parameters parameters are are mostly mostly constant, constant, re representingpresenting the the development development level level of of products products in various industries at the the present present stage. stage. These These consta constants,nts, which which are are shown shown in in Table Table 11,, areare partlypartly ﬁttedfitted fromfrom statistics statistics and partly from actual measurements.

Table 1. The values of parameters correspond to the Mini-Phantom UAV design.

Parameter Value Unit Description

CL 0.8 - Lift coefficient CD 0.04 - Drag coefficient 2 Imax 950 [W/m ] Maximum irradiance ktemp 0.89 Temperature influence factor (5 ◦C) kbat 290 [Wh/kg] Energy density of Li-ion battery 2 ksc 0.29 [kg/m ] Mass density of solar cells 2 kenc 0.13 [kg/m ] Mass density of encapsulation kmppt 0.0042 [kg/W] Mass to power ratio of MPPT kprop 0.008 [kg/W] Mass to power ratio of propulsion group 3 ka f 0.44/9.81 [kg/m ] Structural mass constant mav 0.11 [kg] Mass of autopilot system ηbec 0.8 - Efficiency of step-down converter ηsc 0.2 - Efficiency of solar cells ηcbr 0.9 - Efficiency of the curved solar panels ηcharg 0.95 - Efficiency of battery charge ηctrl 0.95 - Efficiency of motor controller ηdcharg 0.95 - Efficiency of battery discharge ηmot 0.9 - Efficiency of motor ηmppt 0.97 - Efficiency of MPPT ηplr 0.85 - Efficiency of propeller Pav 1.5 [W] Power of autopilot system x1 3.1 - Airframe mass wingspan exponent x 0.25 - Airframe mass aspect ratio exponent 2 − Appl. Sci. 2020, 10, 1300 10 of 26

The second part is the aerodynamic model, including the parameters that reflect the flight environment and aircraft lift and drag characteristics, which are mainly determined by the flight mission objectives. Mission parameters of Mini-Phantom UAV are shown in Table2. In the aerodynamic model, the aircraft layout parameters are represented by aspect ratio and wingspan as input variables.

Table 2. Mission parameters for Mini-Phantom UAV.

Parameter Value Unit Description

mpld 0.1 [kg] Payload mass ηwthr 0.85 - Irradiance margin factor Ppld 2 [W] Payload power consumption ρ 0.78 [kg/m3] Air density (4500 m) Tday 14.3 [h] Summer day duration (34.3◦ N)

The third part is the balance between the energy consumed by the aircraft and the solar energy captured, which connects the weight model with the aerodynamic model, forming a closed loop, through which the optimal aircraft layout can be obtained.

3. Detailed Design

3.1. Layout Design Solar-powered UAVs adopt various forms of layout, such as flying-wing layout, tandem-wing layout, conventional layout, etc. Among them, the flying-wing layout solar-powered UAV is typically representative of NASA’s ‘Pathfinder’ and ‘Helios’. Compared with the conventional layout, the wing layout has the advantages of light structure weight and high aerodynamic efficiency, which are crucial for long-endurance aircraft. However, the static stability of the flying-wing layout is usually weak, and it is more sensitive to the changes of flight altitude and speed. Moreover, the flight angle of attack of the flying-wing layout UAV is less robust, and the pitching moment is easy to diverge in the state of high angle of attack, which is usually solved by split-drag-rudder or differential thrust control. Therefore, the flying-wing layout is usually used for large wingspan solar UAVs, and the takeoff and landing should be carried out in good conditions. The tandem wing layout consists of two wings, the front and rear, and has a higher lift coefficient with the same wingspan. When the aircraft is disturbed, the fuselage has good aeroelastic characteristics compared with the conventional layout. The stall characteristics of this layout are similar to those of the canard layout. In order to ensure good stall characteristics, the front wing stalls before the rear wing stalls, which has good longitudinal safety in flight. However, in the takeoff and landing stage, the rear wing cannot reach the maximum lift point, so good landing site conditions and long takeoff and landing distance are required. The conventional layout facilitates the design of the fuselage and the arrangement of loads and equipment. Moreover, it has the advantages of high robustness and better stability at high angle of attack, and low takeoff and landing speed, which is especially suitable for small solar-powered UAVs with hand-launched takeoff and hard landing. A fully functional prototype was built, named Mini-Phantom Solar (MP Solar). Considering the extreme flight environment (4600 m altitude, 30 C temperature), the MP Solar UAV (Figure6) is − ◦ of a conventional configuration with an aileron-free rudder and an all-moving elevator. It is worth mentioning that the full-motion elevator is designed to facilitate disassembly and transportation, and during the landing phase, the full-moving elevator is deflected at a large degree so that the UAV enters a deep stall before it touches ground, which allows the aircraft to achieve a lower landing speed. Appl. Sci. 2020, 10, x FOR PEER REVIEW 12 of 28

3. Detailed Design

3.1. Layout Design Solar-powered UAVs adopt various forms of layout, such as flying-wing layout, tandem-wing layout, conventional layout, etc. Among them, the flying-wing layout solar-powered UAV is typically representative of NASA’s ‘Pathfinder’ and ‘Helios’. Compared with the conventional layout, the wing layout has the advantages of light structure weight and high aerodynamic efficiency, which are crucial for long-endurance aircraft. However, the static stability of the flying-wing layout is usually weak, and it is more sensitive to the changes of flight altitude and speed. Moreover, the flight angle of attack of the flying-wing layout UAV is less robust, and the pitching moment is easy to diverge in the state of high angle of attack, which is usually solved by split-drag-rudder or differential thrust control. Therefore, the flying-wing layout is usually used for large wingspan solar UAVs, and the takeoff and landing should be carried out in good conditions. The tandem wing layout consists of two wings, the front and rear, and has a higher lift coefficient with the same wingspan. When the aircraft is disturbed, the fuselage has good aeroelastic characteristics compared with the conventional layout. The stall characteristics of this layout are similar to those of the canard layout. In order to ensure good stall characteristics, the front wing stalls before the rear wing stalls, which has good longitudinal safety in flight. However, in the takeoff and landing stage, the rear wing cannot reach the maximum lift point, so good landing site conditions and long takeoff and landing distance are required. The conventional layout facilitates the design of the fuselage and the arrangement of loads and equipment. Moreover, it has the advantages of high robustness and better stability at high angle of attack, and low takeoff and landing speed, which is especially suitable for small solar-powered UAVs with hand-launched takeoff and hard landing. A fully functional prototype was built, named Mini-Phantom Solar (MP Solar). Considering the extreme flight environment (4600 m altitude, −30 °C temperature), the MP Solar UAV (Figure 6) is of a conventional configuration with an aileron-free rudder and an all-moving elevator. It is worth mentioning that the full-motion elevator is designed to facilitate disassembly and transportation, and during the landing phase, the full-moving elevator is deflected at a large degree so that the UAV enters a deep stall before it touches ground, which allows the aircraft to achieve a lower landing speed. Appl. Sci. 2020, 10, 1300 11 of 26

Figure 6. The Mini-Phantom Solar UAV. Dimensions are given in mm.

The main wing, with a 3.2 m wingspan, can be split into three segments of similar size to facilitate transport. The aileron-free design of the rectangular wing ensures the wing’s integrity, which increases the proportion of solar cells laid (14 units of solar cells laid as two rows on each wing segment). The outer wing has a dihedral angle of 9 degrees to improve the lateral stability of the aircraft. The layout characteristics are summarized in Table3.

Table 3. Design characteristics of Mini-Phantom Solar UAV.

Parameter Value Unit Description AR 10.6 - Aspect ratio b 3.2 m Wingspan 2 Swin 0.934 m Wing area 2 Sele 0.06 m Horizontal tail area 2 Svert 0.08 m Vertical tail area 2 Srud 0.017 m Rudder area VH 0.21 - Horizontal tail volume VV 0.28 - Vertical tail volume m 2.9 kg Total mass mbat 1.2 kg Battery mass mMpld 0.6 kg Max payload mass V 10 m/s Flight speed

3.2. Aerodynamic Design The lift–drag ratio of an aircraft is used to evaluate the aerodynamic efficiency and is the key parameter affecting the endurance of solar-powered UAV. The flying-wing layout has a high lift–drag ratio due to the absence of fuselage and tail. In order to improve aerodynamic efficiency, the layout of flying-wings is used for reference in aerodynamic design. The airfoil of the main wing is optimized on the basis of the low Reynolds number laminar airfoil to make its tail have the characteristics of reverse camber (Figure7b), reduce the nose-down pitching moment generated by the main wing, and thus reduce the requirement for the horizontal tail volume (Tables1 and3). The airfoil of tails is NACA0009, which is developed by National Advisory Committee for Aeronautics (NACA). The aerodynamic efficiency of the UAV can be improved by increasing the reverse camber of airfoil, reducing the area of the horizontal tail and the length of the rear fuselage, and the stability of the UAV at high angle of attack can be guaranteed by the existence of the tail. Appl. Sci. 2020, 10, x FOR PEER REVIEW 13 of 28

Figure 6. The Mini-Phantom Solar UAV. Dimensions are given in mm.

Table 3. Design characteristics of Mini-Phantom Solar UAV.

Parameter Value Unit Description AR 10.6 - Aspect ratio b 3.2 m Wingspan 2 Swin 0.934 m Wing area

2 Sele 0.06 m Horizontal tail area

2 Svert 0.08 m Vertical tail area

2 Srud 0.017 m Rudder area

VH 0.21 - Horizontal tail volume

VV 0.28 - Vertical tail volume m 2.9 kg Total mass

mbat 1.2 kg Battery mass

mMpld 0.6 kg Max payload mass V 10 m/s Flight speed

3.2. Aerodynamic Design The lift–drag ratio of an aircraft is used to evaluate the aerodynamic efficiency and is the key parameter affecting the endurance of solar-powered UAV. The flying-wing layout has a high lift–drag ratio due to the absence of fuselage and tail. In order to improve aerodynamic efficiency, the layout of flying-wings is used for reference in aerodynamic design. The airfoil of the main wing is optimized on the basis of the low Reynolds number laminar airfoil to make its tail have the characteristics of reverse camber (Figure 7b), reduce the nose-down pitching moment generated by the main wing, and thus reduce the requirement for the horizontal tail volume (Tables 1 and 3). The airfoil of tails is NACA0009, which is developed by National Advisory Committee for Aeronautics (NACA). The aerodynamic efficiency of the UAV can be improved by increasing the reverse camber of airfoil, reducing the area of the horizontal tail and the length of the rear fuselage, and the stability of theAppl. UAV Sci. 2020at high, 10, 1300angle of attack can be guaranteed by the existence of the tail. 12 of 26

Appl. Sci. 2020, 10, x FOR PEER REVIEW 14 of 28

(a) (b)

Figure 7. (a) Aerodynamic computing grid model. (b) Optimized airfoil of main wing.

Aerodynamic calculations(a) were carried out for the shape of the UAV,(b )among which the turbulence model was the Spalart–Allmaras model, and the calculated state was: Flight altitude was Figure 7. (a) Aerodynamic computing grid model. (b) Optimized airfoil of main wing. 5000 m, speedFigure was 7. 10.5(a) Aerodynamicm/s. computing grid model. (b) Optimized airfoil of main wing. Figure 8 shows the change of aerodynamic parameters of the aircraft with the angle of attack. Aerodynamic calculations were carried out for the shape of the UAV, among which the turbulence When the UAV cruised at an altitude of 5000 m at a speed of 10.5 m/s, the maximum lift-to-drag ratio model was the Spalart–Allmaras model, and the calculated state was: Flight altitude was 5000 m, was 18.4, which was obtained at 2 degrees of angle of attack, which was consistent with the UAV’s speed was 10.5 m/s. trim angle of attack. At the trim angle of attack, the lift coefficient of the UAV was about 0.8 and the Figure8 shows the change of aerodynamic parameters of the aircraft with the angle of attack. stall angle of attack was about 9 degrees. When the UAV cruised at an altitude of 5000 m at a speed of 10.5 m/s, the maximum lift-to-drag ratio It can be seen from the pitching moment curve that the pitching moment converged in the was 18.4, which was obtained at 2 degrees of angle of attack, which was consistent with the UAV’s trim whole process and maintained a good linear relationship. The static stability margin was about angle of attack. At the trim angle of attack, the lift coefficient of the UAV was about 0.8 and the stall 13.8% near the cruising point. With the increase of the angle of attack to more than 6 degrees, the angle of attack was about 9 degrees. pitching moment curve became steeper, which means that the UAV had stronger longitudinal static It can be seen from the pitching moment curve that the pitching moment converged in the whole stability at large angle of attack and also means that the UAV had good takeoff and landing process and maintained a good linear relationship. The static stability margin was about 13.8% near characteristics and stall characteristics. the cruising point. With the increase of the angle of attack to more than 6 degrees, the pitching Through the analysis of aerodynamic data, it can be seen that the method of improving the moment curve became steeper, which means that the UAV had stronger longitudinal static stability at UAV’s aerodynamic efficiency by increasing the reverse camber of airfoil and reducing the large angle of attack and also means that the UAV had good takeoff and landing characteristics and horizontal tail area was effective, which brought good aerodynamic efficiency. stall characteristics. The mature conventional layout ensured the longitudinal stability of the UAV in the high angle Through the analysis of aerodynamic data, it can be seen that the method of improving the UAV’s of attack and adjacent stall state, which can meet the requirements of hand-launched takeoff and aerodynamic efficiency by increasing the reverse camber of airfoil and reducing the horizontal tail area hard landing in the extreme environment without runway. was effective, which brought good aerodynamic efficiency.

(a) (b)

Figure 8. Cont. Appl.Appl. Sci. Sci. 20202020, ,1010,, x 1300 FOR PEER REVIEW 1513 of of 28 26

FigureFigure 8. 8. ComputationalComputational Fluid Fluid Dynamics Dynamics (CFD) (CFD) result result of of cruise cruise at at 5000-m 5000-m altitude altitude of of the the UAV: UAV: ( (aa)) LiftLift coefficient, coefficient, angle angle of of attack attack data; data; ( (bb)) drag drag coefficient, coefficient, angle angle of of attack attack data; data; ( (cc)) pitching pitching moment moment coefficient,coefficient, angle angle of of attack attack data; data; ( d)) lift–drag lift–drag ratio ratio data. data.

3.3. StructureThe mature and conventionalAvionics layout ensured the longitudinal stability of the UAV in the high angle of attack and adjacent stall state, which can meet the requirements of hand-launched takeoff and hard landingThe instructure the extreme of MP environment Solar UAV withoutwas a traditional runway. beam-rib structure made of carbon, KEVLAR, and balsa wood. The beam of the main wing was a rectangular hollow section beam consisting of rectangular3.3. Structure carbon and Avionics tubes arranged up and down and walls with lightening holes. The rectangular beam had better bending resistance than the circular carbon tube beam, which was harder to deform The structure of MP Solar UAV was a traditional beam-rib structure made of carbon, KEVLAR, under overload caused by sudden wind, ensuring the safety of solar cells. The wing ribs, trusses, and and balsa wood. The beam of the main wing was a rectangular hollow section beam consisting of the solar cells laid on the upper surface of the wing formed a shell structure, ensuring good torsional rectangular carbon tubes arranged up and down and walls with lightening holes. The rectangular stiffness of the wing (Figure 9b). beam had better bending resistance than the circular carbon tube beam, which was harder to deform To prevent the propeller from interfering with the optical payload, the fuselage was of a under overload caused by sudden wind, ensuring the safety of solar cells. The wing ribs, trusses, and tail-pushed configuration, in which the propulsion system is located at the rear of the fuselage. The the solar cells laid on the upper surface of the wing formed a shell structure, ensuring good torsional hard landing adopted by MP Solar UAV required the fuselage to have enough strength, so the stiffness of the wing (Figure9b). fuselage structure (Figure 9a) adopted a shell structure composed of transverse mediastinum frame, To prevent the propeller from interfering with the optical payload, the fuselage was of a tail-pushed truss, and skin. configuration, in which the propulsion system is located at the rear of the fuselage. The hard landing adopted by MP Solar UAV required the fuselage to have enough strength, so the fuselage structure (Figure9a) adopted a shell structure composed of transverse mediastinum frame, truss, and skin.

ImageThe Transmission power generation Resiver and storageStick Antenna systemElectronic consisted Governor of solar cells, MPPT, and lithium-ion batteries, GNSS Solar-modules which were the coreRC of Resiver the MP Solar UAV’sMPPT perpetualBrushless Motor flight. The energy was(Monocrystalline generated silion) by 42 Controller of Holder Traling-edge monocrystalline silicon solar cells laid on the surface of the main wing and stored in two high-energy(Carbon fibre ) density batteries located via MPPT. The packaged lithium-ion battery was connected in a 3S (11.1v) layout,Airspeed whichTube can provide 354 Wh energy. Solar cells had an energyLeading-edge conversion efficiency of about 23%. Spar It is important to note that low temperatures can seriously affSquareect batterybeam discharge performance(Balsa wood) (Carbon fibre ) Ribs and thus threaten flight endurance. Thus, the use of low thermal insulationWing materialskin (Balsa turnedwood) the Zoom Camera Telemetry Autopilot Carbon Tail Tube fuselage into a closedLithium-ion cabin, Battery and theLink self-heating from the batteries and avionics kept the cabin at the suitable temperature. Avionics systems (See Figure(a) 9a for avionics system arrangement) of MP Solar(b) UAV included autopilot [Figure14], digital 9. (a)The radio, fuselage remote structure control and (RC) arrangement receiver, of sensors, instruments, propulsion (b) wing systems, structure. etc. In order to obtain necessary information of position velocity and altitude for flight controller, many low-cost andThe highly power robust generation MEMS (micro-electro-mechanical and storage system consisted systems) of sensorssolar cells, were MPPT, equipped and onlithium-ion this UAV, batteries,including which accelerometer were the gyroscope,core of the MP magnetometer, Solar UAV’s barometer, perpetual flight. and GPS. The Theenergy actuation was generated system was by 42driven monocrystalline by two brushless silicon servos solar to drivecells thelaid rudder on the and surface all-moving of the elevator, main wing which and had highstored sensitivity in two high-energyand reliability. density The propulsion batteries located system via used MPPT. a brushless The packaged motor direct lithium-ion drive 10- battery7-inch was propeller, connected with in × Appl. Sci. 2020, 10, x FOR PEER REVIEW 15 of 28

Figure 8. Computational Fluid Dynamics (CFD) result of cruise at 5000-m altitude of the UAV: (a) Lift coefficient, angle of attack data; (b) drag coefficient, angle of attack data; (c) pitching moment coefficient, angle of attack data; (d) lift–drag ratio data.

3.3. Structure and Avionics The structure of MP Solar UAV was a traditional beam-rib structure made of carbon, KEVLAR, and balsa wood. The beam of the main wing was a rectangular hollow section beam consisting of rectangular carbon tubes arranged up and down and walls with lightening holes. The rectangular beam had better bending resistance than the circular carbon tube beam, which was harder to deform under overload caused by sudden wind, ensuring the safety of solar cells. The wing ribs, trusses, and the solar cells laid on the upper surface of the wing formed a shell structure, ensuring good torsional stiffness of the wing (Figure 9b). To prevent the propeller from interfering with the optical payload, the fuselage was of a tail-pushed configuration, in which the propulsion system is located at the rear of the fuselage. The Appl. Sci. 2020, 10, 1300 14 of 26 hard landing adopted by MP Solar UAV required the fuselage to have enough strength, so the Appl. Sci. 2020, 10, x FOR PEER REVIEW 16 of 28 fuselage structure (Figure 9a) adopted a shell structure composed of transverse mediastinum frame, atruss,a maximum 3S and(11.1v) skin. power layout, output which of can about provide 430 W 354 at takeo Wh ffenergy.. The UAV Solar off eredcells ahad fully an manual energy mode conversion to deal withefficiency a severe of caseabout of 23%. autopilot failure [15]. It is important to note that low temperatures can seriously affect battery discharge performance and thus threaten flight endurance. Thus, the use of low thermal insulation material turned the Image Transmission Resiver Stick Antenna Electronic Governor GNSS Solar-modules fuselage into a closedRC cabin, Resiver and the self-heatingMPPT frBrushlessom the Motor batteries and avionics kept(Monocrystalline the cabin silion) at the Controller of Holder suitable temperature. Traling-edge (Carbon fibre ) Avionics systems (See Figure 9a for avionics system arrangement) of MP Solar UAV included autopilot [14], digital radio, remote control (RC) receiver, sensors, propulsion systems, etc. In order Airspeed Tube Leading-edge to obtain necessary information of position velocity and altitude for flight controller, many low-costSpar Square beam (Balsa wood) and highly robust MEMS (micro-electro-mechanical systems) sensors(Carbon fibre were ) equipped onRibs this UAV, Wing skin (Balsa wood) Zoom Camera Telemetry Autopilot including accelerometer gyroscope, magnetometer,Carbon Tail Tubebarometer, and GPS. The actuation system was Lithium-ion Battery Link driven by two brushless servos to drive the rudder and all-moving elevator, which had high sensitivity and reliability. The(a) propulsion system used a brushless motor direct(b) drive 10- × 7-inch propeller, with a maximum power output of about 430 W at takeoff. The UAV offered a fully manual FigureFigure 9. (a)The fuselage structure structure and and arrangement arrangement of of instruments, instruments, (b (b) )wing wing structure. structure. mode to deal with a severe case of autopilot failure [15]. MP Solar UAV was designed to have a maximum payload capacity of 450 g. Different MPThe Solarpower UAV generation was designed and storage to have asystem maximum consisted payload of capacitysolar cells, of 450MPPT, g. Di andfferent lithium-ion functional functional payloads mean different payload weights, and the aircraft can be fitted with different payloadsbatteries, meanwhich diwerefferent the payloadcore of the weights, MP Solar and UA theV’s aircraft perpetual can flight. be fitted The with energy diff waserent generated loads to by suit loads to suit different missions. The aircraft had a designed payload capacity of 50 g for perpetual di42ff erentmonocrystalline missions. The silicon aircraft solar had cells a designed laid on payloadthe surface capacity of the of 50main g for wing perpetual and stored flight, whichin two is flight, which is enough to carry a small fixed-focus camera with video recording capability. The enoughhigh-energy to carry density a small batteries fixed-focus located camera via MPPT. with The video packaged recording lithium-ion capability. battery The UAV was connected can also carry in UAV can also carry a 450 g optical-infrared optoelectronic pod for complex missions at the expense a 450 g optical-infrared optoelectronic pod for complex missions at the expense of perpetual flight of perpetual flight capability. The whole energy route of payload and avionics instruments are capability. The whole energy route of payload and avionics instruments are shown in Figure 10. shown in Figure 10.

FigureFigure 10.10. Topological system overview over the the MP MP Solar Solar UAV. UAV. Appl. Sci. 2020, 10, 1300 15 of 26

4. UAV Flight Control

4.1. Lateral-Directional Nonlinear Dynamic Model The wings of a solar-powered UAV provide the main solar cell laying surface, so a complete wing without split by control surface is especially important for a small solar-powered UAV. In this paper, the MP Solar UAV was aileron-free and used rudder to control both yaw axis and roll axis. The rudder deﬂection generated yaw moment, and the sideslip angle generated by the rudder deﬂection also brought a large roll moment because the wing had a dihedral design [16]. The lateral-directional nonlinear dynamic model [17] of aileron-free UAV was established as Formula (19).

.  = + +  v pw ru g cos θ sin ϕ Y/m  . −  ϕ = p + q sin ϕ tan θ + r cos ϕ tan θ  .  = + (19)  ψ q sin ϕ sec θ r cos ϕ sec θ  .  p = Γ1pq Γ2qr + Γ3l + Γ4n  . −  r = Γ pq Γ qr + Γ l + Γ n 5 − 1 4 6 where the speciﬁc expansion of symbols are shown in the Formula (20).

 2  Γ = IxIz Ixz  −   Γ1 = Ixz Ix Iy + Iz /Γ  h − i  Γ = I I I + I2 /Γ  2 z z y xz  −  Γ3 = Iz/Γ   Γ = I /Γ  4 xz  2 (20)  Γ5 = Ix Ix Iy + Ixz  −  =  Γ6 Ix/Γ  qbS  Y = C p + C r + qS C β + C r  2V Yp Yr Yβ Yδr  qb2S  l = C p + C r + qbS C + C  2V lp lr lββ lδr δr  2  = qb S + + +  n 2V Cnpp Cnrr qbS Cnββ Cnδr δr where Ix, Iy, Iz, and Ixz are the moment of inertia and the product of inertia of the UAV, respectively. The u,v, and w are the velocity component of ﬂight path under the body axes. The p, q, and r are the angular rate of the body axes. The ϕ, θ, and ψ are angle of roll, pitch, and yaw, respectively. Y, l, n are lateral force, rolling moment, and yawing moment. The q is dynamic pressure and δr is the angle of rudder deﬂection.

4.2. Analysis of Stability and Flight Quality By using Taylor expansion, the nonlinear dynamics model is linearized and the lateral-directional small-disturbance equations are obtained as Formula (21).

. x = Ax + Bu (21)

h iT where state vector x = β p r ϕ ψ , state matrix A is shown in Formula (22), and B is control T matrix, control input u = [δr] .    Yβ α +YP Yr 1 g cos θ /V 0   ∗ − ∗ ∗   L L L 0 0   β p r    A =  Nβ Np Nr 0 0  (22)    0 1 tan θ 0 0   ∗   0 0 sec θ 0 0  ∗ Appl. Sci. 2020, 10, x FOR PEER REVIEW 18 of 28

αθ∗+ − ∗ ∗ YYYgVβ 1cos 0 Pr LLβ pr L 00  A = NN N 00 β pr (22) 01tan0θ ∗ 0 Appl. Sci. 2020, 10, 1300 16 of 26 θ ∗ 00sec0 0

wherewhereα αand∗ Vandrepresent V ∗ represent angle of angle attack of and attack speed and of speed UAV atoftrim UAV point. at trim The point. characteristic The characteristic roots of ∗ ∗ lateral-directionalroots of lateral-directiona state matrixl state represent matrix represent the ﬂight the quality flight of quality UAV lateral-directional of UAV lateral-directional mode. mode. TheThe designdesign missionmission environmentenvironment ofof MPMP SolarSolar UAVUAV waswas forfor 4500-m4500-m elevationelevation aboveabove plateau.plateau. Therefore,Therefore, thethe aileron-free UAV UAV still still needs needs to to have have a good a good flight ﬂight performance performance in high in high altitude. altitude. The Thechange change of theof theUAV’s UAV’s system system characteristic characteristic roots roots with with speed speed increase increase at an at altitude an altitude of 500 of m 500 and m and5000 5000m, respectively, m, respectively, are shown are shown in Figure in Figure 11. The11. Thespeed speed range range was was8–12 8–12 m/s. m /s.

(a)

Appl. Sci. 2020, 10, x FOR PEER REVIEW 19 of 28

(b)

(c)

FigureFigure 11. TheThe change change of of system system characteristic characteristic roots roots with with different diﬀerent altitude altitude and and speed speed in in lateral-directionallateral-directional mode. mode. ( (aa)) Spiral Spiral mode. mode. ( (bb)) Roll Roll mode. mode. (c (c) )Dutch Dutch roll roll mode. mode.

It can be seen from Figure 11a,b that, when flying at high altitude, the characteristic root of roll mode and spiral mode moved to the right of the complex plane compared with that at low altitude, and the flight quality became worse. The move of characteristic roots of the Dutch roll mode (Figure 11c) in real axis had no obvious directionality, but the characteristic roots dispersion was larger than that of low-altitude flight, which means that the stability of the Dutch roll mode was more sensitive to the flight speed at high altitude. At the same altitude, with the increase of flight speed, the characteristic roots of lateral-directional mode all moved to the left of the complex plane, and the flight quality became better. The Cooper–Harper rating is a systematic method of quantifying the results of an experiment, which is divided into three grades to describe the flight quality of an aircraft [18]. Among them, the level-one flight quality represents the flight quality to ensure the successful completion of all flight missions. At an altitude of 500 m, both the roll mode and the Dutch roll mode were of level-one flight quality in the full speed range. However, when the speed was lower than 9 m/s, the characteristic root of spiral mode was positive and the doubling time was 23 s. In other words, although the spiral mode diverged at a low speed, it still met the level-one flight quality requirements. At an altitude of 5000 m, the maximum rolling constant of convergence time was 0.13. With the change of flight speed, the frequency of Dutch rolling mode almost remained unchanged, about 2.32, and the minimum damping was about 0.149. That is, both the rolling mode and Dutch rolling mode (Figure 11b,c) met the level-one quality within the full speed range of 5000 m. At the cruise speed of 10 m/s, the half-value period was 25.7 s, and the spiral mode met the level-one flight quality requirements. However, when the flight speed was lower than 9 m/s, the spiral mode no longer met the level-one flight quality, and continued to deteriorate with the flight speed decrease. Therefore, high-altitude flight needs to avoid flying at too low a speed to prevent deterioration of flight quality.

4.3. Analysis of Lateral-Directional Maneuverability The UAVs with conventional configuration use aileron and rudder to control roll and sideslip, while MP Solar UAV with aileron-free design uses rudder to controls both roll and sideslip. Therefore, it is necessary to analyze the maneuvering characteristics of aileron-free UAV [19]. Figure 12 shows the free response of rudder deflection during cruise at altitudes of 500 m and 5000 m. Appl. Sci. 2020, 10, 1300 17 of 26

It can be seen from Figure 11a,b that, when flying at high altitude, the characteristic root of roll mode and spiral mode moved to the right of the complex plane compared with that at low altitude, and the flight quality became worse. The move of characteristic roots of the Dutch roll mode (Figure 11c) in real axis had no obvious directionality, but the characteristic roots dispersion was larger than that of low-altitude flight, which means that the stability of the Dutch roll mode was more sensitive to the flight speed at high altitude. At the same altitude, with the increase of flight speed, the characteristic roots of lateral-directional mode all moved to the left of the complex plane, and the flight quality became better. The Cooper–Harper rating is a systematic method of quantifying the results of an experiment, which is divided into three grades to describe the flight quality of an aircraft [18]. Among them, the level-one flight quality represents the flight quality to ensure the successful completion of all flight missions. At an altitude of 500 m, both the roll mode and the Dutch roll mode were of level-one flight quality in the full speed range. However, when the speed was lower than 9 m/s, the characteristic root of spiral mode was positive and the doubling time was 23 s. In other words, although the spiral mode diverged at a low speed, it still met the level-one flight quality requirements. At an altitude of 5000 m, the maximum rolling constant of convergence time was 0.13. With the change of flight speed, the frequency of Dutch rolling mode almost remained unchanged, about 2.32, and the minimum damping was about 0.149. That is, both the rolling mode and Dutch rolling mode (Figure 11b,c) met the level-one quality within the full speed range of 5000 m. At the cruise speed of 10 m/s, the half-value period was 25.7 s, and the spiral mode met the level-one flight quality requirements. However, when the flight speed was lower than 9 m/s, the spiral mode no longer met the level-one flight quality, and continued to deteriorate with the flight speed decrease. Therefore, high-altitude flight needs to avoid flying at too low a speed to prevent deterioration of flight quality.

4.3. Analysis of Lateral-Directional Maneuverability The UAVs with conventional configuration use aileron and rudder to control roll and sideslip, while MP Solar UAV with aileron-free design uses rudder to controls both roll and sideslip. Therefore, it is necessary to analyze the maneuvering characteristics of aileron-free UAV [19]. Figure 12 shows the free response of rudder deflection during cruise at altitudes of 500 m and 5000 m. According to Figure 12, the rudder deflection of the UAV can cause both sideslip angle and roll angle, scilicet, and the rudder can be used for rolling and yaw control. There was an obvious phase difference between the roll angular rate and the sideslip angle, which indicates that the roll angle of the UAV was mainly caused by the sideslip angle, rather than the rudder deflection. Compared with aileron control roll, rudder control is an indirect process with slower response, which needs to be considered when designing the control laws. Compared with the flight at low altitude, the response frequency of the UAV at high altitude was almost the same, but the amplitude was larger, which is consistent with the result in the Figure 11 that the system damping decreased as the altitude increased. After the convergence of shock, the effect of rudder control sideslip at different altitudes was basically the same, while the effect of high-altitude control was better than that of low-altitude control. Appl.Appl. Sci. Sci.2020 2020,,10 10,, 1300 x FOR PEER REVIEW 2018 of of 28 26

(a) (b)

FigureFigure 12.12. Free response of of rudder rudder deflection deflection at at low low altitude altitude and and high high altitude: altitude: (a) (Ruddera) Rudder deflect deflect 20 20degree. degree. (b) ( brudder) rudder deflect deflect 1 degree. 1 degree. 4.4. Successive Loop Closure Flight Control According to Figure 12, the rudder deflection of the UAV can cause both sideslip angle and roll 4.4.1.angle, Longitudinal scilicet, and Controlthe rudder Law can be used for rolling and yaw control. There was an obvious phase difference between the roll angular rate and the sideslip angle, which indicates that the roll angle of the UAVThe mission was mainly environment caused by of the MP sideslip Solar UAVangle, was rather cruising than the ata rudder speed deflection. of 10 m/s on Compared a plateau with at an altitudeaileron ofcontrol 4500 m.roll, The rudder longitudinal control is control an indirect input process of the UAVwith wasslower all-moving response, elevator which needs and throttle, to be andconsidered the longitudinal when designing dynamic the state control space, laws. which is shown in Formula (23), was extended to consider the influenceCompared of elevatorwith the and flight throttle. at low altitude, the response frequency of the UAV at high altitude was almost the same, but the amplitude. was larger, which is consistent with the result in the Figure xlon = Alonxlon + Blonulon 11 that the system damping decreased as theh altitude increased.iT After the convergence of shock, the(23) effect of rudder control sideslip at differentxlon = alutitudes w q wasθ basically the same, while the effect of high-altitude control was better than that of low-altitude control. Where u, w, q, and θ correspond to x-, z- axis velocity component, pitch angular rate, and pitch angle.4.4. Successive The control Loop law Closure of longitudinal Flight Control inner loop was altitude hold (Formula (24)), and the output was command signal of deflection angle of all-moving elevator and throttle. 4.4.1. Longitudinal Control Law qc = kq(θc θ) − (24) The mission environment of MP= Solar ( + UAV 1was+ cruising)( at a) speed of 10 m/s on a plateau at an δe kp kpi s kpd s qc q altitude of 4500 m. The longitudinal control input· of the· UAV −was all-moving elevator and throttle, and Thethe longitudinal outer loop was dynamic the height state andspace, airspeed which is hold shown control in Formula law, which (23), are was shown extended in Formula to consider (25), andthe theinfluence simulation of elevator results and of thethrottle. UAV’s pitch angle and altitude command tracking are shown in the Figure 13. Both the pitch angle and the altitude =+ can well realize the command tracking without static xAxBulon lon lon lon lon error or overshoot, and the response time of the pitch angleT command was about 1.1 s. (23) = θ xlon uwq ki θc = k (hc h) + h (hc h) Where u, w, q, and θ correspond to x-, z- paxish velocity component,s pitch angular rate, and pitch angle. − k − (25) c iV c The control law of longitudinal innerδt = kloopp (V wasV aaltitude) + ( Vhold V(Formulaa) (24)), and the output was V a − s a − command signal of deflection angle of all-moving elevator and throttle. Appl. Sci. 2020, 10, x FOR PEER REVIEW 21 of 28

=−θθ qkcqc() (24) δ =+⋅+⋅1 − epp()()kk ksqq p c ids The outer loop was the height and airspeed hold control law, which are shown in Formula (25), and the simulation results of the UAV’s pitch angle and altitude command tracking are shown in the Figure 13. Both the pitch angle and the altitude can well realize the command tracking without static Appl.error Sci. 2020 or overshoot,, 10, 1300 and the response time of the pitch angle command was about 1.1 s. 19 of 26

(a) (b)

FigureFigure 13. 13.Simulation Simulation of pitch angle angle and and altitude altitude command command tracking: tracking: (a) Pitch (a) angle Pitch tracking. angle tracking. (b) (b)Altitude tracking. tracking.

4.4.2. Lateral-Directional Control Law The lateral-directional control law is still based on the successive closed-loop theory [20]. For the

ki UAV with aileron-free design, the rudderθ cc=−+− was the onlyh cinput for lateral-directional control. That is, khp ()() h h h h s the rudder can simultaneously control the two channels of roll and sideslip: Channel 1, the(25) rudder ki was treated as a pseudo-aileron, controllingδ =−+− thecc roll channel.V Formula (26) is the control law when the tpaakV()() V V aa V rudder is used as a pseudo-aileron. ChannelV 2, the rudders was used to control the sideslip of the UAV, and Formula (27) is the output of rudder command during coordinated turn. 4.4.2. Lateral-Directional Control Law pc = (kp + kd s)(φc φ) The lateral-directional control law is stillφ basedφ · on the− successive closed-loop theory [20]. For(26) δa = (kp + kp s)(pc p) the UAV with aileron-free design, the rudder was thed · only −input for lateral-directional control. That is, the rudder can simultaneously control the two gchannels of roll and sideslip: Channel 1, the rudder rc = tan φc was treated as a pseudo-aileron, controlling the Vroll0 channel. Formula (26) is the control law when(27) δr = kr(rc r) the rudder is used as a pseudo-aileron. Channel 2, the− rudder was used to control the sideslip of the UAV, and Formula (27) is the output of rudder command during coordinated turn. The outer loop was a straight-line trajectory tracking by the L1 method [21], Figure 14 shows the path diagram of UAV and linear target.=+⋅− In the figure,φφd is track error, L1 is the distance between pkkscpd()()φφ c conference point and UAV, V is velocity component of the horizontal plane, χ is the flight path angle,(26) as δ =+⋅()()kkspp − appd c is expected lateral acceleration, and ψt is the azimuth angle of target trajectory. The control instruction of horizontal trajectory tracking based on L1 method is shown in Formula (28). g r = tanφ cc . V2 V0 V V (27) as = 2 sin η 2 d + d L1 δ =−kr()L r1 L1 rrc≈ . V V (28) φcmd = arctan 2 L g d + L d The outer loop was a straight-line trajectory tracking1 by the1 L1 method [21], Figure 14 shows the = + path diagram of UAV and linear ψtarget.cmd Inψ tthe ηfigure,1 d is track error, L1 is the distance between conference point and UAV, V is velocity component of the horizontal plane, χ is the flight path The simulation results of UAV straight-line trajectory tracking are shown in the Figure 15. Figure 15a shows that the response time of the roll angle command was about 1.4s and there was no static error or overshoot. Figure 15b shows that the response time of lateral distance command (50 m) was about 10.39 s with 2.12 m overshoot. The simulation results show that the trajectory tracking control based on L1 method is feasible for the UAV flight at 4500-m altitude with aileron-free configuration. Appl. Sci. 2020, 10, x FOR PEER REVIEW 22 of 28

angle, as is expected lateral acceleration, and ψt is the azimuth angle of target trajectory. The control instruction of horizontal trajectory tracking based on L1 method is shown in Formula (28).

Appl. Sci. 2020, 10, x FOR PEER REVIEW 22 of 28 angle,Appl. Sci. as2020 is expected, 10, 1300 lateral acceleration, and ψt is the azimuth angle of target trajectory. The control20 of 26 instruction of horizontalFigure trajectory 14. Schematic tracking diagram based ofon UAV L1 method and desired is shown flight inpath. Formula (28).

2  =≈+VVVη  adds 2sin2 LLL111 VV (28) φ =+arctan 2 dd cmd  Lg11 L ψψη=+ cmd t 1 The simulation results of UAV straight-line trajectory tracking are shown in the Figure 15. Figure 15a shows that the response time of the roll angle command was about 1.4s and there was no static error or overshoot. Figure 15b shows that the response time of lateral distance command (50 m) was about 10.39 s with 2.12 m overshoot. The simulation results show that the trajectory tracking control based on L1 method is feasible for the UAV flight at 4500-m altitude with aileron-free configuration. FigureFigure 14. 14. SchematicSchematic diagram diagram of of UAV UAV and and desired desired flight ﬂight path. path.

2  =≈+VVVη  adds 2sin2 LLL111 VV (28) φ =+arctan 2 dd cmd  Lg11 L ψψη=+ cmd t 1 The simulation results of UAV straight-line trajectory tracking are shown in the Figure 15. Figure 15a shows that the response time of the roll angle command was about 1.4s and there was no static error or overshoot. Figure 15b shows that the response time of lateral distance command (50 m) was about 10.39 s with 2.12 m overshoot. The simulation results show that the trajectory tracking (a) (b) control based on L1 method is feasible for the UAV flight at 4500-m altitude with aileron-free configuration.Figure 15. Simulation of straight-lin straight-linee trajectory tracking: ( a)) Roll Roll angle angle command tracking, ( b) lateral distance command tracking.

5. Field Experiment

5.1. Experiment Environment 5.1. Experiment Environment The target mission area of the MP Solar UAV was Qiangtang and Hoh Xil in Tibet (low temperature The target mission area of the MP Solar UAV was Qiangtang and Hoh Xil in Tibet (low and high altitude), where it is harsh for human activities. The purpose of the experiment was to verify temperature and high altitude), where it is harsh for human activities. The purpose of the whether the scheme had the ability to fly in extreme environments and to analyze the feasibility of perpetual flight in the plateau region. Full-system flight tests of MP Solar UAV were performed in extreme environments, as shown in the Figure 16. The experiment site is located at 4604 m in Qiangtang, (31.988◦ E, 87.317◦ N) and the experiment was conducted in winter with a temperature of 30 C. Vehicle-mounted takeoff was − ◦ adopted because the human body could not withstand strenuous activity at high altitude. The whole (a) (b) process of flight experiments included hand-launched takeoff in a car, autonomous climb, cruise, and slideFigure [22]. 15. Simulation of straight-line trajectory tracking: (a) Roll angle command tracking, (b) lateral distance command tracking.

5. Field Experiment

5.1. Experiment Environment The target mission area of the MP Solar UAV was Qiangtang and Hoh Xil in Tibet (low temperature and high altitude), where it is harsh for human activities. The purpose of the Appl. Sci. 2020, 10, x FOR PEER REVIEW 23 of 28 Appl. Sci. 2020, 10, x FOR PEER REVIEW 23 of 28 experiment was to verify whether the scheme had the ability to fly in extreme environments and to experimentanalyze the wasfeasibility to verify of perpetuawhether lthe flight scheme in the had plateau the ability region. to fly in extreme environments and to analyzeFull-system the feasibility flight of tests perpetua of MPl Solarflight UAV in the were plateau performed region. in extreme environments, as shown in the FigureFull-system 16. The flight experiment tests of MP site Solar is located UAV atwere4604 performed m in Qiangtang, in extreme (31.988 environments,° E, 87.317° as N) shown and the in theexperiment Figure 16. was The conducted experiment in sitewinter is located with a at temperature 4604 m in Qiangtang, of −30 °C. (31.988Vehicle-mounted° E, 87.317° takeoffN) and wasthe experimentadopted because was conducted the human inbody winter could with not awithstand temperature strenuous of −30 activity°C. Vehicle-mounted at high altitude. takeoff The whole was adoptedprocess ofbecause flight theexperiments human body incl coulduded nothand-launched withstand strenuous takeoff in activity a car, atautonomous high altitude. climb, The wholecruise, processand slide of [22]. flight experiments included hand-launched takeoff in a car, autonomous climb, cruise, Appl.and Sci. slide2020 ,[22].10, 1300 21 of 26

(a) (b) (c) (a) (b) (c) Figure 16. The high-altitude field experiment of UAV in Qiangtang, Tibet: (a) Vehicle-mounted FigureFiguretakeoff, 16. 16. (Theb ) Thecruise high-altitude high-altitude phase for field mission, field experiment experiment (c) hard oflanding UAVof UAV inin Qiangtang,extreme in Qiangtang, environment. Tibet: Tibet: (a) Vehicle-mounted (a) Vehicle-mounted takeo ff, (b)takeoff, cruise phase(b) cruise for phase mission, for (mission,c) hard landing(c) hard inlanding extreme in extreme environment. environment. 5.2. Flight Parameters of HighAltitude Experiment 5.2.5.2. Flight Flight Parameters Parameters of of HighAltitude HighAltitude ExperimentExperiment The flight parameters of takeoff and cruise at high altitude are shown in Figure 17. Figure 17a The flight parameters of takeoff and cruise at high altitude are shown in Figure 17. Figure 17a showsThe the flight climbing parameters process of of takeoff the UAV: and Thecruise UAV at hi tookgh altitude off from are an shown altitude in ofFigure 4600 17.m, Figureclimbed 17a to showsshows4700 the m, the climbingand climbing turned process process to level of theof flight. the UAV: UAV: In The order The UAV toUAV tookensure took off thefrom off safety from an altitude anof takingaltitude of 4600off of at4600 m, high climbed m, altitude, climbed to 4700 the to m, and4700vehicle-mounted turned m, and to level turned flight.takeoff to level In was order flight. adopted, to ensureIn order and the tothe safety ensure instantaneous of takingthe safety off airspeedat of high taking altitude, of off departure at thehigh vehicle-mounted altitude,was 16 m/s.the takeovehicle-mountedFigureff was 17b adopted, presents takeoff anda part thewas of instantaneous parameters adopted, and from airspeed the the instantaneous stable of departurecruise phase.airspeed was The 16 of absolute m departure/s. Figure cruising was17b presents16altitude m/s. a partFigureof ofthe parameters UAV17b presents was from4700 a partm, the the stableof parametersaltitude cruise was phase. from about the The 100 stable absolute m, cruisecruise cruising airspeedphase. altitudeThe was absolute about of the 10.5cruising UAV m/s, was altitudeand 4700 the m, theofflight altitude the trajectoryUAV was was about was4700 100a m,circular m,the cruise altitude route airspeed around was about wasthe mission about100 m, 10.5 cruise area. m / s,airspeed and the was flight about trajectory 10.5 m/s, was and a circular the routeflight around trajectory the missionwas a circular area. route around the mission area.

4705 10 H 4703 command command 5 4705 H 10 4701 H 4703 command command 50 4699 H 4701 -5 4697 0 4699 4695 -10-5 4697 16 30 V 469520 -1014 command 16 30 command V 10 12 V 200 14 command command 10 V 10-10 12 -200 8 10 -10-30 6 -20400 168 -30350 146 400300 16 350250 12 14 300200 150 10 250 12 100 200 8 15050 10 1000 6 0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 8 0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 50 0 6 0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 (a) (b) (a) (b) FigureFigure 17. 17.Flight Flight result result of of fieldfield experiment:experiment: ( aa)) Climb Climb phase, phase, (b (b) )cruise cruise phase. phase. Figure 17. Flight result of field experiment: (a) Climb phase, (b) cruise phase. InIn this this field field experiment, experiment, the the UAV’s UAV’s missionmission flightflight path mainly mainly included included a aquadrilateral quadrilateral flight flight In this field experiment, the UAV’s mission flight path mainly included a quadrilateral flight pathpath and and a a circular circular flight flight path.path. The The quadrilateral quadrilateral flight flight path path was was suitable suitable for takeoff for takeo andff landingand landing and path and a circular flight path. The quadrilateral flight path was suitable for takeoff and landing and andsearching searching targets targets in ina awide wide range. range. The The circular circular flight flight path path was was used used for for tracking tracking and and monitoring monitoring searching targets in a wide range. The circular flight path was used for tracking and monitoring designateddesignated areas areas or or targets. targets. Figure Figure 18 18 shows shows the the circular circular flight flight path path and and the the quadrilateralquadrilateral flightflight pathpath in designated areas or targets. Figure 18 shows the circular flight path and the quadrilateral flight path thein flight the flight test, respectively.test, respectively. in the flight test, respectively. From the circular flight path (Figure 18a), it can be seen that the aileron-free UAV tracked the path well under the condition of no wind, and the flight trajectory had almost no overshoot. Figure 18b shows the situation of the UAV tracking a quadrilateral flight path with 7 m/s positive wind. The UAV tracked the path well on the upwind side. However, on the downwind side, the flight path had overshoot. In general, the aileron-free UAV achieved the required path tracking for the mission, although maybe the tracking was less accurate than aileron UAV in windy conditions. In order to increase the proportion of solar cells laid and reduce the structural weight of the wings brought by the control surface, the UAV eliminated ailerons as roll control and used a rudder to control both roll axis and yaw axis. Figure 12 shows the free response simulation of rudder deflection when the UAV flew at different altitudes. In the flight test, the control characteristics of the UAV were verified. Appl. Sci. 2020, 10, x FOR PEER REVIEW 24 of 28

From the circular flight path (Figure 18a), it can be seen that the aileron-free UAV tracked the path well under the condition of no wind, and the flight trajectory had almost no overshoot. Figure 18b shows the situation of the UAV tracking a quadrilateral flight path with 7 m/s positive wind. The UAVAppl. Sci. tracked 2020, 10 the, x FOR path PEER well REVIEW on the upwind side. However, on the downwind side, the flight path24 hadof 28 overshoot. In general, the aileron-free UAV achieved the required path tracking for the mission, althoughFrom maybe the circular the tracking flight waspath less (Figure accurate 18a), than it can aileron be seen UAV that in the windy aileron-free conditions. UAV tracked the path well under the condition of no wind, and the flight trajectory had almost no overshoot. Figure 18b shows the situation of the UAV tracking a quadrilateral flight path with 7 m/s positive wind. The UAV tracked the path well on the upwind side. However, on the downwind side, the flight path had overshoot. In general, the aileron-free UAV achieved the required path tracking for the mission, Appl.although Sci. 2020 maybe, 10, 1300 the tracking was less accurate than aileron UAV in windy conditions. 22 of 26

(a) (b)

Figure 18. The flight path of the solar-powered UAV: (a) Circular flight path without wind, (b) quadrilateral flight path with 7 m/s wind.

In order to increase the proportion of solar cells laid and reduce the structural weight of the wings brought by the control(a) surface, the UAV eliminated ailerons as roll control(b) and used a rudder to controlFigure both 18.18. TheTheroll flight flightaxis pathand path ofyaw of the the axis. solar-powered Figure 12 UAV: UAV:shows (a ) the Circular free flightflightresponse path simulation without wind, of rudder( b) deflectionquadrilateral when the flightflight UAV pathpath flew withwith 7at7 mm/s different/s wind.wind. altitudes. In the flight test, the control characteristics of the UAV were verified. FigureFigureIn order 1919 toshowsshows increase the the data datathe proportionof of the the roll roll angle angle of solar and and cellrudder rudders laid deflection deflection and reduce when when the the thestructural UAV UAV changed changed weight from fromof the a a circularcircularwings brought flightflight path pathby the to to acontrol straight-linea straight-line surface, flight the flight path,UAV path, whichelim inatedwhich means aileronsmeans the UAV asthe roll turned UAV control leftturned fromand leftused the clockwisefrom a rudder the clockwisecircularto control route circular both and roll route entered axis and and a straight-lineentered yaw axis. a straight-line Figure flight path. 12 fshowslight Asseen path. the in Asfree the seen figure,response in th thee simulationfigure, roll angle the commandrollof rudder angle commandappeareddeflection appearedatwhen 15 s the with at UAV 15 a commands flewwith ata command different value of altitudes.value 30 degrees, of 30 In degrees, the and flight the and rudder test, the the rudder deflected control deflected characteristics after 0.3 after s and 0.3 ofs a anddeflectionthe UAVa deflection were value verified. ofvalue 28 degrees of 28 degr afterees 1.5 after s. The 1.5 roll s. The angle roll response angle response time was time slightly was slower,slightly about slower, 5 s, aboutbecauseFigure 5 s, the because roll19 shows angle the the of roll the data angle aileron-free of the of rollthe UAV angleaileron- was andfree not rudder UAV directly deflection was caused not directlywhen by the the rudder caused UAV deflection changedby the rudder from but by a deflectionthecircular sideslip, flight but which by path the was sideslip,to ana straight-line indirect which process was flight an forindirect path, the rudder. processwhich In meansfor general, the rudder.the the UAV response In general,turned characteristics leftthe responsefrom the of characteristicstheclockwise roll angle circular wereof the route consistent roll angleand entered with were the consistent a simulation straight-line with results. theflight simulation path. As results. seen in the figure, the roll angle command appeared at 15 s with a command value of 30 degrees, and the rudder deflected after 0.3 s and a deflection value of 28 degrees after 1.5 s. The roll angle response time was slightly slower, about 5 s, because the roll angle of the aileron-free UAV was not directly caused by the rudder deflection but by the sideslip, which was an indirect process for the rudder. In general, the response characteristics of the roll angle were consistent with the simulation results.

FigureFigure 19. 19. Rudder—rollRudder—roll angle angle response. response.

According to different mission objectives, the UAV can carry different weight and power consumption of the payloads. Weight and power consumption and other parameters will affect the weight coefficient of batteries. Heavier and higher power consumption of payload usually means worse flight endurance. In the fieldFigure test, 19. the Rudder—roll UAV completed angle response. its flight test with two different payloads: Payload 1 was a small fixed-focus camera weighing only 50 g and a low-power transmission with power of 600 mW and payload 2 was an optical-infrared composite optronics pod weighting 450 g and a remote digital graph transmission with an effective distance of 15 km and a power of 15 W. In the case of constant total weight, different payloads not only mean different additional power consumption, but also affect the carrying capacity of batteries, which is closely related to flight endurance. Figure 20 shows different payloads applied to endangered animal monitoring. Appl.Appl. Sci.Sci. 20202020,, 1010,, xx FORFOR PEER REVIEW 2525 of of 28 28

AccordingAccording to different mission objectives,objectives, ththee UAVUAV cancan carrycarry differentdifferent weight weight and and power power consumptionconsumption of the payloads. Weight andand powerpower coconsumptionnsumption and and other other parameters parameters will will affect affect the the weightweight coefficientcoefficient of batteries. Heavier andand higherhigher powerpower consumptionconsumption of of payload payload usually usually means means worseworse flightflight endurance. In the field test,test, thethe UAUAVV completedcompleted itsits flightflight testtest with with two two different different payloads:payloads: PayloadPayload 1 was a small fixed-focusfixed-focus cameracamera weighingweighing onlyonly 5050 gg andand a a low-powerlow-power transmissiontransmission withwith power of 600 mWmW andand payloadpayload 22 was was an an optical-infrared optical-infrared composite composite optronics optronics pod pod weightingweighting 450450 g and a remote digital graphgraph transmissitransmissionon with with an an effective effective distance distance of of 15 15 km km and and a a powerpower ofof 1515 W. In the case of constant totaltotal weight,weight, differentdifferent payloadspayloads not not only only mean mean different different additionaladditional powerpower consumption, but alsoalso affectaffect thethe carryingcarrying capacitycapacity of of batteries, batteries, which which is is closely closely relatedrelated toto flightflight endurance. Figure 2020 showsshows didifferentfferent payloadspayloads appliedapplied toto endangered endangered animal animal Appl.monitoring.monitoring. Sci. 2020, 10, 1300 23 of 26

(a) (b) (a) (b) Figure 20. Different payloads for endangered animals: (a) Optical image, (b) infrared image. FigureFigure 20.20. DiDifferentfferent payloadspayloads forfor endangeredendangered animals:animals: ((aa)) OpticalOptical image,image, ((bb)) infraredinfrared image.image. The Figure 21 shows the cruising power of the UAV with two different payloads. The dotted The Figure 21 shows the cruising power of the UAV with two different payloads. The dotted line inThe the Figure figure 21 is showsthe average the cruising cruising power power, of which the UAV can bewith used two to differentestimate payloads.the aircraft’s The battery dotted line in the figure is the average cruising power, which can be used to estimate the aircraft’s battery lineendurance in the figureto determine is the average whether cruising the UAVpower, is whichcapable can of be flying used toacross estimate the thenight, aircraft’s which battery is a endurance to determine whether the UAV is capable of flying across the night, which is a prerequisite enduranceprerequisite to of determineperpetual flight. whether the UAV is capable of flying across the night, which is a ofprerequisite perpetual flight.of perpetual flight.

(a) (b) (a) (b) FigureFigure 21.21. CruiseCruise power data of the UAV UAV with different different payload payload configurations configurations at at altitude altitude of of 4700 4700 m: m: (Figurea(a)) Small Small 21. fixed-focus Cruisefixed-focus power camera datacamera with of the low-powerwith UAV low-power with transmission, different transmission, payload (b) optical-infrared configurations (b) optical-infrared combined at altitude optoelectronic combined of 4700 m: pod(optoelectronica) withSmall remote fixed-focus pod digital with transmission.recameramote digital with transmission.low-power transmission, (b) optical-infrared combined optoelectronic pod with remote digital transmission. DiDifferentfferent seasons seasons have have di ffdifferenterent hours hours of sunshine, of sunshine, which which is closely is closely related torelated the flight to the endurance. flight Asendurance. canDifferent be seen As seasons fromcan be the seen have Figure from different 21 thea, theFigure hours cruising 21a, of the powersuns cruisinghine, of the powerwhich UAV of withis theclosely payloadUAV relatedwith 1 waspayload to about the 1 27wasflight W, whichendurance.about means27 W, As which the can battery bemeans seen endurance the from battery the ofFigure endurance the UAV 21a, wastheof the cruising 13.3 UAV h, whichwas power 13.3 was of h, the enoughwhich UAV was towith allowenough payload the to UAV allow 1 was to crossaboutthe UAV the 27 nightW, to crosswhich (perpetual the means night flight) the (perpetual battery in summer. enduranceflight) In in summer su ofmmer. the (takeUAV In summer July was 1 13.3 as (take an h, example), which July 1 wasas thean enough flightexample), time to allow the that thetheflight solarUAV time cellto that cross power the the solar can night meet cell (perpetual power the cruise can flight) consumptionmeet inthe su cruisemmer. is consumption about In summer 11.7 h, (take is and about theJuly night11.7 1 as h, an time and example), isthe 12.3 night h the to simplifyflight time the that neglect the solar of transition cell power time. can In meet winter the (takecruise December consumption 10 as is an about example), 11.7 h, the and maximum the night solar irradiance and sunshine duration are shorter, and considering the influence of low temperature in winter on battery discharge efficiency, the flight endurance that the solar cell power can meet the cruise consumption is only 6.1 h, which means the UAV cannot cross the night. If the UAV takes off at midnight, it reaches its energy balance point when batteries run out and the UAV will gain its maximum endurance of 29.7 h. As can be seen from the Figure 21b, the cruising power of the UAV with payload 1 was about 39 W, which means the battery endurance of the UAV was 9.2 h and the UAV cannot cross the night with payload 2. The maximum endurance in summer and winter are 29.3 h and 20.8 h, respectively. On the whole, at an altitude of 4600 m, the MP Solar UAV can achieve autonomous climb, cruise, hover, and descent control, completing the feasibility verification of the whole process of control system Appl. Sci. 2020, 10, 1300 24 of 26 design. The cruise power data of the UAV with different loads indicate that it is feasible for the UAV to achieve perpetual flight at an altitude of 4600 m, which verifies the correctness of the design method. The permanent flight of solar UAV requires both seasons and mission loads. Less ideal seasons and more stringent mission requirements will prevent the UAV from achieving perpetual flight, but it still has better endurance, which is the unique advantage of solar-powered UAV.

6. Conclusions This paper proposed a complete research process of a hand-launched solar-powered UAV for plateau, from conceptual design method, detailed design, dynamics modeling, control scheme design, numerical simulation, and hardware in the loop simulation experiment to field experiments. In order to achieve perpetual flight in the plateau region, this paper proposed a conceptual design method of solar-powered UAV based on energy balance and the optimal parameter solution was found for the plateau region as the mission environment. The main conceptual parameters indicate that it is feasible for small, solar-powered UAV to fly day and night (perpetual flight) at high altitudes during the summer. In the design model, the solar irradiance model took latitude, altitude, date, and other parameters as input, which can more accurately reflect the sunshine conditions at different geographical locations and time compared with the simplified model. In this paper, real irradiance data were collected by the total irradiance meter to modify the model and the main parameters of the modified model, such as maximum illumination and sunshine duration, were in good agreement with the experimental data. In view of the low-temperature flight environment of the UAV, the temperature influence factor was introduced into the traditional battery model, and the battery weight model considering the low-temperature influence was proposed. Among them, the temperature influence factor was measured by the low-temperature discharge test of the battery. To validate the design methodology, a full-system, hand-launched, solar-powered UAV was designed and fabricated, including configuration, structure, payload selection, and whole avionics. Based on the six-degrees-of-freedom equation of motion, this paper analyzed the changes in the stability and maneuverability of aileron-free UAV at high altitude. At high altitude, the stability of both spiral mode and roll mode were worse, and the speed sensitivity of Dutch roll mode increased. At high altitude, the damping of UAV was reduced and the oscillation convergence after disturbance was slower. According to the flight conditions of the plateau, the longitudinal height tracking control law with continuous closed loop and the lateral-directional trajectory tracking control law based on L1 method were designed, respectively. Finally, the flight test was conducted at an altitude of 4600 m in Qiangtang (31.988◦ E, 87.317◦ N), Tibet, and the feasibility of the design method and control system was verified. The cruise power data of the UAV with different loads indicate that it is feasible for the UAV to achieve perpetual flight at an altitude of 4600 m during summer. Less ideal seasons and more stringent mission requirements will prevent the UAV from achieving perpetual flight, but it still has better endurance, which is the unique advantage of solar-powered UAV. The design method and verification method can be extended to other similar UAV platform.

Author Contributions: X.Z. (Xin Zhao) and Z.Z. conducted the whole design method and system modeling. X.Z. (Xin Zhao) and A.G. contributed the designing and manufacture of hardware system of prototype. X.Z. (Xiaoping Zhu) analyzed the experiment result. X.Z. (Xin Zhao) and Z.Z. wrote the paper. All authors have read and agreed to the published version of the manuscript. Funding: This research was funded by the Innovation plan of the Institute grant number TC2018DYDS24 and Key R&D Projects in Shaanxi Province of China grant number 2018ZDCXL-GY-03-04. Conﬂicts of Interest: The authors declare no conﬂict of interest. Appl. Sci. 2020, 10, 1300 25 of 26

References

1. Colella, N.; Wenneker, G. Pathfinder: Developing a solar rechargeable aircraft. IEEE Potentials 1996, 15, 18–23. [CrossRef] 2. Helios Prototype Solar-Powered Aircraft. Available online: https://www.nasa.gov/centers/dryden/history/ pastprojects/Helios/index.html (accessed on 20 December 2019). 3. Solar Impulse 2 Detailed Route. Available online: https://zh.m.wikipedia.org/zh-cn/SolarImpulse2 (accessed on 6 January 2020). 4. Ackerman, E. Giant Solar-Powered UAVs are Atmospheric Satellites. Available online: https://spectrum. ieee.org/automaton/robotics/drones/giant-solar-powered-uavs-are-atmospheric-satellites (accessed on 6 January 2020). 5. QinetiQ, QinetiQ files for three world records for its Zephyr Solar powered UAV. Asia-Pac. Def. Rep. 2010, 36, 8. 6. Airbus Zephyr S Breaks World Flight Endurance Record during Maiden Flight. Available online: https://www.sae.org/news/2018/08/airbus-zephyr-s-breaks-world-flight-endurance-record-during- maiden-flight (accessed on 6 January 2020). 7. Noth, A. Design of Solar Powered Airplanes for Continuous Flight. Ph.D. Thesis, Swiss Federal Institute of Technology (ETH), Zürich, Switzerland, September 2008. 8. Oettershagen, P.; Stastny, T.; Hinzmann, T. Robotic technologies for solar-powered UAVs: Fully autonomous updraft-aware aerial sensing for multiday search-and-rescue missions. J. Field Robot. 2017, 35, 1–29. [CrossRef] 9. Keidel, B. Auslegung und Simulation von Hochfliegenden Dauerhaft Stationierbaren Solardrohnen. Ph.D. Thesis, Technischen Universität München, München, Germany, 2000. 10. Badescu, V.; Gueymard, C.A.; Cheval, S.; Oprea, C.; Baciu, M.; Dumitrescu, A.; Iacobescu, F.; Milos, I.; Rada, C. Computing global and diffuse solar hourly irradiation on clear sky. Review and testing of 54 models. J. Renew. Sustain. Energy Rev. 2012, 16, 1636–1656. [CrossRef] 11. Romeo, G.; Frulla, G.; Cestino, E.; Corsino, G. Heliplat: Design, Aerodynamic, Structural Analysis of Long-Endurance Solar-Powered Stratospheric Platform. J. J. Aircr. 2004, 41, 1505–1520. [CrossRef] 12. Youngblood, J.W.; Talay, T.A.; Pegg, R.J. Design of Long-Endurance Unmanned Airplanes Incorporating Solar and Fuel Cell Propulsion. In Proceedings of the The 20th Joint Propulsion Conference, AIAA-84-1430, Cincinnati, OH, USA, 11–13 June 1984. 13. Rizzo, E.; Frediani, A. A Model for Solar Powered Aircraft Preliminary Design. In Proceedings of the International Conference on Computational & Experimental Engineering & Sciences 04, Madeira, Portugal, 26–29 July 2004; Volume 1, pp. 39–54. 14. Ebeid, E.; Skriver, M.; Terkildsen, K.H.; Jensen, K.; Schultz, U.P. A Survey of Open-Source UAV Flight Controllers and Flight Simulators. Microprocess. Microsyst. 2018, 61, 11–20. [CrossRef] 15. Arifianto, O.; Farhood, M. Development and Modeling of a Low-Cost Unmanned Aerial Vehicle Research Platform. J. Intell. Robot. Syst. 2015, 80, 139–164. [CrossRef] 16. Cheng, X.; Wang, H.P.; Zhang, Y.Z. Electric and no aileron SUAV’s lateral-direct dynamic characters. Flight Dyn. 2009, 27, 74–77. (In Chinese) 17. Beard, R.W.; Mclain, T.W. Small Unmanned Aircraft: Theory and Practice; Princeton University Press: Princeton, NJ, USA, 2012; pp. 52–92. 18. Bandu, N.P. Performance, Stability, Dynamics, and Control of Airplanes; American Institute of Aeronautics and Astronautics: Reston, VA, USA, 2013; pp. 511–514. (In Chinese) 19. Ma, Z.Y.; Zhu, X.P.; Zhou, Z. A lateral-directional control method combining rudder and propeller for full-wing solar-powered UAV. J. Acta Aeronaut. Et Astronaut. Sin. 2018, 39, 321633. (In Chinese) 20. Ma, Z.Y.; Zhu, X.P.; Zhou, Z. On-ground lateral direction control for an unswept fly-wing UAV. Aeronaut. J. 2019, 123, 416–432. [CrossRef] Appl. Sci. 2020, 10, 1300 26 of 26

21. Park, S.; Deyst, J.; How, J.P. A new nonlinear guidance logic for trajectory tracking. In Proceedings of the AIAA Guidance, Navigation, and Control Conference an Exhibit, Reston, VA, USA, 16–19 August 2004. 22. Guo, A.; Zhou, Z.; Zhu, X.; Bai, F. Low-Cost Sensors State Estimation Algorithm for a Small Hand-Launched Solar-Powered UAV. Sensors 2019, 19, 4627. [CrossRef][PubMed]

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).