Control of a Dual-Stator Flux-Modulated Motor for Electric Vehicles

energies

Article Control of a Dual-Stator Flux-Modulated Motor for Electric Vehicles

Xinhua Guo 1,*, Shaozhe Wu 1,2, Weinong Fu 2, Yulong Liu 2, Yunchong Wang 2 and Peihuang Zeng 1

1 College of Information Science and Engineering, Huaqiao University, Xiamen 361021, China; [email protected] (S.W.); [email protected] (P.Z.) 2 Department of Electrical Engineering, The Hong Kong Polytechnic University, Kowloon 999077, Hong Kong, China; [email protected] (W.F.); [email protected] (Y.L.); [email protected] (Y.W.) * Correspondence: [email protected]; Tel.: +86-159-8583-3956

Academic Editor: K. T. Chau Received: 29 April 2016; Accepted: 28 June 2016; Published: 2 July 2016

Abstract: This paper presents the control strategies for a novel dual-stator ﬂux-modulated (DSFM) motor for application in electric vehicles (EVs). The DSFM motor can be applied to EVs because of its simple winding structure, high reliability, and its use of two stators and rotating modulation steels in the air gap. Moreover, it outperforms conventional brushless doubly-fed machines in terms of control performance. Two stator-current-oriented vector controls with different excitation in the primary winding, direct and alternating current excitation, are designed, simulated, and evaluated on a custom-made DSFM prototype allowing the decoupled control of torque. The stable speed response and available current characteristics strongly validate the feasibility of the two control methods. Furthermore, the proposed control methods can be employed in other applications of ﬂux-modulated motors.

Keywords: dual-stator; ﬂux-modulated; vector control (VC); electric vehicles (EVs); decouple control

1. Introduction With the emergence of an energy crisis, ways of reducing air-pollution have become the great challenge [1]. Electric vehicles (EVs) contribute significantly to energy saving and environmental protection, and on account of these benefits, they constitute today’s direction for the automotive industry [2]. A compact and highly-reliable electric traction motor, which has high efficiency over a wide speed range with load-adaptive controllable speed-torque contours, is crucial for any electric propulsion system [3,4]. Traction motors of EVs should have high torque density in terms of weight and volume, a wide range of torque-speed characteristics, high operating efficiency, and high reliability. The novel structure of a brushless, doubly-fed generator (BDFG) proposed in [5], which has the aforementioned advantages, can be efficiently applied to EVs. To distinguish the different applications of such machines, the concept of dual-stator flux-modulated (DSFM) motors for EVs was presented. Figure1 illustrates a novel machine, which has a three-phase, 36-slot, 12-pole winding on the outer stator and three-phase, 24-slot, 10-pole winding on the inner stator. Compared with most permanent magnet synchronous motors (PMSMs), the DSFM motor with a wide range of speeds and torques and no flux-weakening control yields higher performance in EVs [5].

Energies 2016, 9, 517; doi:10.3390/en9070517 www.mdpi.com/journal/energies Energies 2016, 9, 517 2 of 20 Energies 2016, 9, 517 2 of 19 Outer winding Energies 2016, 9, 517 Outer stator 2 of 20

Outer winding Outer stator

Iron segImroenn ts segmenItsnner Inner Isntanteor r Innewr inding stator winding (a) (b) (a) (b)

FigureFigure 1. Structure 1. Structure of the of DSFMthe DSFM motor motor with with an an iron iron segment segment rotor: ( a(a) )3D 3D model model; and; and (b) (2Db) 2Dmodel model. . Figure 1. Structure of the DSFM motor with an iron segment rotor: (a) 3D model; and (b) 2D model. ConventionalConventional brushless brushless, doubly, doubly-fed-fed machines machines (BDFMs), (BDFMs), such such as as BDFG, BDFG, have have been been designed designed and investigatedConventionaland investigated only brushless, only for forwind doubly-fedwind energy energy systems machines systems [ 6[6– (BDFMs),–1010].]. InIn particular, such as BDFG,control control havestrategies strategies been for designed forBDF BDFMs, andMs, suchinvestigated such as direct as only direct torque for torque wind and and energy flux flux control systems control [ 11 [ [611],–10], voltage]. voltage In particular, vector vector--orientedoriented control control strategies control (VC) (VC) for [1 BDFMs,2, [1132],, 13 and], such and fieldas direct-orientedfield torque-oriented control and control ﬂux (FOC) control (FOC) [1 4 [11 [1,154],,]15, voltage ]have, have been vector-oriented been investigated. investigated. control The The FOC(VC) FOC and[ 12 and, 13 VC ], VC and schemes schemes ﬁeld-oriented were were comparedcontrolcompared (FOC) in [1 [ 514in] ,and[1155],] andthe have theadvantages been advantages investigated. and and disadvantages disadvantages The FOC and ofof the VC twotwo schemes co controlntrol weremethods methods compared were were presented. in presented. [15] and The structural diagram of the brushless, doubly-fed reluctance machine (BDFRM (G)) drive setup is Thethe advantagesstructural diagram and disadvantages of the brushless of the, twodoubly control-fed methodsreluctance were machine presented. (BDFRM The structural(G)) drive diagramsetup is depicted in Figure 2 [13]. depictedof the brushless, in Figure doubly-fed 2 [13]. reluctance machine (BDFRM (G)) drive setup is depicted in Figure2[ 13].

(2Va-Vb-Vc)/3=Vα ~ DC Link+ (2Va-V(bV-Vb-cV)/c)3/=√3V=αVβ ωp ωs Brake chopper ~ DC Link+ (Vb-Vc)/√3=Vβ ωp Shaft ωs Brake chopper Sensor isa isb i Shaft ab iα=ia SV- SV-PWM Vector Vabc Sensor isa isb 1/U iβ=(ia+2ib)/√3 PWM Controller iab iα=ia Vabc Pr θr isα isβ S V- SV-PWM Vector Vα Vβ θrm -jθ 1/U iβ=(ia+2ib3)//√2 3 PeWM Controller Frame Angle θp θ Pr θr × si sα isβ Calcs：A ro B θrm -jθ Vα Vβ 3/2 e isd isq * V Frame Angle Vαβ θp θs i sd × × PI sd × Calcs：A roiα B=ia Q=⅔(iα*uβ-iβ*uα) * PI -jθ iissdq isq e 2/3 i =(i +2i )/√3 P=⅔(iα*uα-iβ*uβ) × PI × PI β Va αβb * Vsd * i sd iαβ P × PI × PI iα=ia Q=⅔(iα*u*β-iβ*uα) * Vsq -jθ Q i sq e 2/3 iβ=(ia+2ib)/√3 P=⅔(iα*u*α-iβ*uβ) ×× PIPI × PI *ω m ωm iαβ P * Vsq Q * × PI ω m ωm Figure 2. A structural diagram of the BDFRM (G) drive setup.

However, theFigure application 2. A structural of the flux diagram-modulated of the motor BDFRM to EVs (G) drivehas not setup. been fully investigated, Figure 2. A structural diagram of the BDFRM (G) drive setup. especially the control strategies. In this paper, the air gap structures of the DSFM motor and BDFRM were compared and dynamic models were investigated, and the operating principle of this motor However, the application of the ﬂux-modulatedflux-modulated motor to EVs has not been fully investigated investigated,, was verified through the open-loop variable voltage and variable frequency (VVVF) control. especially the control strategies strategies.. In this paper, t thehe air gap structures of the DSFM motor and BDFRM were comparedMoreover, andtwo dynamicclosed loop models control were methods, investigated, stator-current and the-oriented operating vector principle control (SCOVC) of this motor with was were compareddirect current and (DC) dynamic excitation model and alternatings were investigated, current (AC) and excitation, the operating were proposed principle to ensure of this that motor veriﬁed through the open-loop variable voltage and variable frequency (VVVF) control. Moreover, two was vtheerified motor through operates the at different open-loop speeds variable and load voltage conditions. and variableSimulation frequency and experiment (VVVF) results control. closed loop control methods, stator-current-oriented vector control (SCOVC) with direct current (DC) Moreover,obtained two for closed a customised loop control prototype methods, completely stator validated-current the- feasibilityoriented of vector the two control control (SCOVC)strategies. with directexcitation current and (DC) alternating excitation current and (AC) alternating excitation, current were (AC) proposed excitation, to ensure were that proposed the motor to ensure operates that at thedifferent motor2. Dynamic speeds operates andModelling at load different conditions. of Dual speeds-Stator Simulation Flux and- Modulated load and conditions. experiment (DSFM) SimulationMotors results obtained and experiment for a customised results prototype completely validated the feasibility of the two control strategies. obtained forFigure a customised 3 presents prototype the air gap completely structures invalidated the BDFRM the feasibilityand DSFM of motor the two with control an ideal strategies. stator surface around the rotor. The air gaps in the two motors were assumed to be uniform, implying that 2. Dynamic Modelling of Dual-Stator Flux-Modulated (DSFM) Motors 2. Dynamicthe air Modellinggap length between of Dual the-Stator stator Flux and- Modulatedthe rotor teeth (DSFM) is ge and Motors that between the stator and the Figurerotor yoke3 presents is ge + hm the in Figure air gap 3a, structures that the air ingap the length BDFRM between and the DSFM two stator motor and with the iron an segments ideal stator Figure 3 presents the air gap structures in the BDFRM and DSFM motor with an ideal stator surface around the rotor. The air gaps in the two motors were assumed to be uniform, implying that surface around the rotor. The air gaps in the two motors were assumed to be uniform, implying that the air gap length between the stator and the rotor teeth is ge and that between the stator and the rotor the air gap length between the stator and the rotor teeth is ge and that between the stator and the yoke is ge + hm in Figure3a, that the air gap length between the two stator and the iron segments on rotor yoke is ge + hm in Figure 3a, that the air gap length between the two stator and the iron segments

Energies 2016, 9, 517 3 of 19 Energies 2016, 9, 517 3 of 20 Energies 2016, 9, 517 3 of 20 theon rotor the rotor is go +is ggio, + and gi, and that that between between the the outer outer stator stator and and the the inner inner stator stator isisg goo + gii ++ hhmm inin Figure Figure 3b3.b. on the rotor is go + gi, and that between the outer stator and the inner stator is go + gi + hm in Figure 3b. BasedBased on on the the preceding preceding analyses, analyses, the the air air gap gap functions functions ofof thethe twotwo motors are illustrated illustrated in in Figure Figure 4.4. Based on the preceding analyses, the air gap functions of the two motors are illustrated in Figure 4.

Outer stator Outer stator Stator core core Stator core core go 0 π go 0 π ge 0 π φ ge 0 π αpm Rotor αpm φ hm αpm Rotor αpm φ h steels hm αpm αpm φ m steels hm αpm αpm Rotor core gi Rotor core gi Inner stator Inner stator Steel Steel core Steel Steel core (a) (b) (a) (b) Figure 3. Air gap structures of the (a) BDFRM motor and (b) DSFM motor. FigureFigure 3. 3Air. Air gap gap structures structures ofof thethe ((aa)) BDFRMBDFRM motor and and ( (bb)) DSFM DSFM motor motor..

g g g g

hm hm hm hm

go+gi ge go+gi ge θrm θrm θrm θrm (a) (b) (a) (b) Figure 4. Air gap function of the (a) BDFRM motor and (b) DSFM motor. Figure 4. Air gap function of the (a) BDFRM motor and (b) DSFM motor. Figure 4. Air gap function of the (a) BDFRM motor and (b) DSFM motor. According to Figure 4a,b, the air gap function of the DSFM motor and that of the BDFRM are According to Figure 4a,b, the air gap function of the DSFM motor and that of the BDFRM are both periodic rectangular waves. The fundamental waves of the two air gap functions can be bothAccording periodic to rectangular Figure4a,b, wave the airs. gap The function fundamental of the waves DSFM ofmotor the and two that air ofgap the functions BDFRM are can both be expressed as periodic sinusoidal waves using Fourier transform [16]. Therefore, in the simplified periodicexpressed rectangular as periodic waves. sinusoid Theal fundamental waves using waves Fourier of transform the two air [16]. gap Therefore, functions in can the be simplified expressed discussion on dynamic model of the DSFM motor, the inverse machine air gap function can be asdiscussion periodic sinusoidal on dynamic waves model using of the Fourier DSFM transform motor, the [16 inverse]. Therefore, machine in the air simpliﬁedgap function discussion can be modelled as follows [17]: onmodelled dynamic as model follows of [1 the7]: DSFM motor, the inverse machine air gap function can be modelled as follows [17]: 1 g1 ( ,  )  m  n cos p (  ) (1) ´g1 ( , rm )  m  n cos p r (  rm ) (1) g pθ, θrmrmq “ m ` ncosprpθ r´ θrmq rm (1) where pr is the number of rotor poles, θ is the mechanical angle around the periphery of the where pr is the number of rotor poles, θ is the mechanical angle around the periphery of the where pr is the number of rotor poles, θ is the mechanical angle around the periphery of the machine machine with respect to the primary A phase axis, and θrm is the rotor mechanical angle with machine with respect to the primary A phaseθ axis, and θrm is the rotor mechanical angle with withreference respect to to the the primary primary A Aphase phase axis. axis, The and m andrm isn thein E rotorquation mechanical (1) are two angle constants with referencein the Fourier to the reference to the primary A phase axis. The m and n in Equation (1) are two constants in the Fourier primaryseries, Aand phase they axis.differ The fromm andthe twon in different Equation air (1) gap are structures, two constants which in mean the Fouriers that the series, two andmotors they series, and they differ from the two different air gap structures, which means that the two motors differhave from similar the air two gap different functions air regardless gap structures, of the which constants. means that the two motors have similar air gap have similar air gap functions regardless of the constants. functionsAccording regardless to of the the first constants. principle-based rigorous development of dynamic equations for the According to the first principle-based rigorous development of dynamic equations for the BDFRM,According the equati to theons ﬁrst for principle-based the primary and rigorous secondary development windings of of dynamic the DSFM equations motor in for a thestationary BDFRM, BDFRM, the equations for the primary and secondary windings of the DSFM motor in a stationary thereference equations frame for the(denoted primary with and the secondary sub-subscript windings s) are [1 of7] the: DSFM motor in a stationary reference reference frame (denoted with the sub-subscript s) are [17]: frame (denoted with the sub-subscript s) are [17]: d d ps  R i λ ps  j   ps Rpp i p s d ps  j   p s υ pspp p sdt ω λ p s ps “ Rpips ` dt  const` j p ps dtθp constp  p const dλ ˇ υ ss ω λ ss “ Rsiss ` ddtsˇ ` j s ss d ˇsθs const (2)  R i ss  j   ssss s s˚ jθ s s (2) s λRss i“ sL i `ˇ L i e r j   s sps sp pdts ˇ ps ss s (2) dt s const ˇ s const˚ jθ λ “ Lsi ` Lpsi e r ss ss * jprs  L i  L i* e jr ps Lp i p s  L ps i s s e psp p s ps s s where the primary winding is the winding of outer* stator,jr and the secondary winding is the winding  L i  L i* e jr ss Ls i s s  L ps i p s e of inner stator in this paper. sss s s ps p s whereThe torque the primary expression winding in the is stationary the winding reference of outer frame stator, is [ 17 and]: the secondary winding is the where the primary winding is the winding of outer stator, and the secondary winding is the winding of inner stator in this paper. winding of inner stator in this paper. 3 The torque expression in theT “stationaryj p L referencei˚ i˚ ejθr frame´ i i ise [1´j7θ]r: (3) The torque expression in thee stationary4 r ps referenceps ss frameps siss [17]: ´ ¯

Energies 2016, 9, 517 4 of 19

Two dq reference frames are needed to convert the equations in the stationary frame to the corresponding rotating frame equations because the frequencies of the two windings in the DSFM motor are different. Assuming that the angle of the primary reference frame is θ, the angle of the secondary reference frame is θr´θ, the conversion formulas can be written as:

λ λ ´jθ pr “ ps e υ υ ´jθ pr “ ps e ´jθ ipr “ ips e ˚ jpθr´θq is “ is e r s (4) ´jpθr´θq λs “ λs e rr s υ υ ´jpθr´θq sr “ ss e ´jpθr´θq isr “ iss e ˚ jθ ipr “ ips e

Substituting the above equations into Equationr (2), the dynamic model of the DSFM motor in the rotating dpqp and dsqs reference frames through strict deductions can be represented as:

dλp υp “ Rpip ` dt ` jωλp dλs υs “ Rsis ` dt ` j pωr ´ ωq λs

λp “ Lp ipd ` jipq ` Lps isd ´ jisq (5) ´ ¯ i˚ ip ` s ˘ looooomooooon λs “ Ls looooomooooonisd ` jisq ` Lps ipd ´ jipq

` is ˘ ´ ˚ ¯ ip looooomooooon looooomooooon The torque expression is: 3 Lps Te “ pr λpdisq ` λpqisd (6) 2 Lp ´ ¯ 3. Experiments Conducted Using the Variable Voltage and Variable Frequency (VVVF) Control The prototype of the DSFM motor was developed experimentally; the test machine and its controller are shown in Figure5. The DSFM motor speciﬁcations are summarized in Table1. The back-to-back controller is comprised of two converters for simultaneously controlling the outer and innerEnergies windings. 2016, 9, 517 5 of 20

(a) (b)

Figure 5 5.. PrototypePrototype ofof the the DSFM DSFM motor motor and and its controller: ( (aa)) t testest machine; and ( b) back-to-backback-to-back controller.

Table 1. DSFM motor prototype parameters and ratings.

Parameter Value Inner winding pole pairs 5 Outer winding pole pairs 6 Rotor pole pairs 11 Inner resistance (Ω) 0.08 Outer resistance (Ω) 0.1 Outer inductance (mH) 4 Inner inductance (mH) 3 Mutual inductance (mH) 0.8 Rated speed (rpm) 924 Maximum speed (rpm) 1300

The VVVF control was implemented to verify the operating principles of the DSFM motor, and the result is shown in Figure 6. The rotor speed was 212 rpm when the frequencies of the outer and inner winding current were fp = 20 Hz and fs = 20 Hz, respectively. Several frequency supplies were then tested using the VVVF control. The results are summarized in Table 2.

The line voltage of A and B phase The outer winding current The inner winding current 100 ) A /

V 50 (

e d u t

i 0 l p m

A -50

-100

0 0.02 0.04 0.06 0.08 0.1 Time (s) Figure 6. Experimental results of the VVVF control.

Energies 2016, 9, 517 5 of 20

(a) (b)

EnergiesFigure2016, 59., Prototype 517 of the DSFM motor and its controller: (a) test machine; and (b) back-to-back controller.5 of 19

Table 1. DSFM motor prototype parameters and ratings. Table 1. DSFM motor prototype parameters and ratings. Parameter Value Inner windingParameter pole pairs Value5 OuterInner w windinginding polepole pairs pairs 6 5 OuterRotor winding pole polepairs pairs 11 6 InnerRotor resistance pole pairs (Ω) 0.08 11 OuterInner resistance resistance ( Ω(Ω)) 0.080.1 Outer resistance (Ω) 0.1 OuterOuter inductance inductance (mH)(mH) 4 InnerInner inductance inductance (mH)(mH) 3 MutualMutual inductance inductance (mH)(mH) 0.8 RatedRated speed speed (rpm)(rpm) 924 Maximum speed (rpm) 1300 Maximum speed (rpm) 1300

The VVVF control was implemented to verify the operating principles of the DSFM motor, and thethe resultresult isis shownshown inin FigureFigure6 6.. The The rotor rotor speed speed was was 212 212 rpm rpm when when the the frequencies frequencies ofof the the outer outer and and innerinner windingwinding currentcurrent werewere ffpp = 20 Hz and ffss = 20 Hz, respectively. Several frequency supplies were thenthen testedtested usingusing thethe VVVFVVVF control.control. TheThe resultsresults areare summarizedsummarized inin TableTable2 .2.

The line voltage of A and B phase The outer winding current The inner winding current 100 ) A /

V 50 (

e d u t

i 0 l p m

A -50

-100

0 0.02 0.04 0.06 0.08 0.1 Time (s) Figure 6.6. Experimental results of the VVVF control.

Table 2. Test results of the VVVF control.

Frequency (Hz) Theoretical Speed (rpm) Real Speed (rpm)

fp = fs = 9 98 96 fp = fs = 20 218 212 fp = fs = 54 589 560

fp = fs = 72 785 748 fp = fs = 85 927 883 fp = fs = 89 970 924

From the test results shown in Table2, one can get the relation among the frequencies of the two windings and the rotor speed: nr “ 60 fp ` fs {pr (7) The differences between the theoretical and real` speeds˘ listed in Table2 were due to the inaccurate speed control of the open loop VVVF control. Energies 2016, 9, 517 6 of 20

Table 2. Test results of the VVVF control.

Frequency (Hz) Theoretical Speed (rpm) Real Speed (rpm) fp = fs = 9 98 96 fp = fs = 20 218 212 fp = fs = 54 589 560 fp = fs = 72 785 748 fp = fs = 85 927 883 fp = fs = 89 970 924

From the test results shown in Table 2, one can get the relation among the frequencies of the two windings and the rotor speed:

nr60 f p f s p r (7)

EnergiesThe2016 differences, 9, 517 between the theoretical and real speeds listed in Table 2 were due to6 of the 19 inaccurate speed control of the open loop VVVF control. 4. Stator-Current-Oriented Vector Control (SCOVC) 4. Stator-Current-Oriented Vector Control (SCOVC) Two additional formulas can be deduced from Equation (7). Two additional formulas can be deduced from Equation (7).

ωrm “ ωp ` ωs {pr (8) rm   p   s p r (8) ` ˘ where ω is the angular frequency of the primary winding current, ω is the angular frequency of the where ωp is the angular frequency of the primary winding current, ωs is the angular frequency of the secondary winding current, and ω is the mechanical angular velocity of the rotor. secondary winding current, and ωrmrm is the mechanical angular velocity of the rotor.

θr r“ θ p ` p θs  s ((9)9) where θθpp,, θθs,s, and andθ rθarer are the the electric electric angles angles of the of primary the primary winding winding current, current, secondary secondary winding winding current, current,and rotor, and respectively. rotor, respectively. According to thethe analysisanalysis ofof thethe vectorvector relationship relationship between between the the two two stator stator windings windings and and rotor rotor in inthe the DSFM DSFM motor, motor, the the vector vector diagram diagram of of the the DSFM DSFM motor motor can can be be representedrepresented asas shownshown in Figure 77[ [1177].].

q s  qp

p

s is  d p

    ds r

 p  r   s 

FigureFigure 7. Reference 7. Reference frames frames and and current current vectors vectors used used in DSFM in DSFM motor motor equations. equations.

The decoupled controlcontrol ofof thethe DSFMDSFM motor motor orients orients the the primary primary winding winding current current vector vectorip toip to the thedp daxis,p axis, which which implies: implies: θ “ θp (10)    p (10) ω “ ωp (11)    ωr ´ ω “ ωr ´ ωp p “ ωs ((12)11) where ωr = pr ˆ ωrm, and: θs “ θr ´ θp (13)

Thus, the secondary winding current vector ip can be automatically oriented to the ds axis. The decoupled voltage equations of the DSFM motor become:

dλp υ “ Rpi ` ` jωpλ p p dt p (14) dλs υs “ Rsis ` dt ` jωsλs

The primary equations are obviously in the ωp frame, and the secondary ones are in the ωs frame. Therefore, this selection moves each of the equations into their “natural” reference frames where the respective vector components appear as DC in the steady state. This decoupled control method proposed in this paper is called SCOVC. Based on the preceding basic analysis of the dynamic model and vector relationship of the DSFM motor, two SCOVC methods with different excitation were investigated and implemented as follows. Energies 2016, 9, 517 7 of 19

4.1. SCOVC with DC Excitation

4.1.1. Theoretical Analysis

If ωp in Equation (11) is equal to 0, which implies that a DC current is inputted in the primary winding, the angle of the primary reference frame θ is equal to 0, then according to Equation (4), the conversion formulas of the primary frame become:

υp “ υp r s (15) ipr “ ips

* Meanwhile, ipq = 0 control was implemented to realize θp = θ = 0. Substituting this value into Equation (13) yields: θs “ θr (16) and the conversion formulas of the secondary frame become:

υ “ υ e´jθr sr ss (17) ´jθr isr “ iss e

That is, a stationary dq reference frame for the primary equation and a rotor reference frame for the secondary equation. The SCOVC method with DC excitation is simply called DC excitation control Energiesin the following2016, 9, 517 text. The DC excitation control method for the DSFM motor is illustrated in Figure8 of 280.

ipα iA iB ipβ Clarke iC ipd * upd i pd PI * ipq upq SVPWM IGBT i pq=0 PI Current Vector outer Oriented Control stator * usd rotor i s PI usα * PI MTPA isd InvPark inner ω r usq usβ SVPWM IGBT ωr PI stator isq isα outer ia Park Clarke isβ stator ib θr ic K

Figure 8. DC excitation control method. Figure 8. DC excitation control method.

WithWith the direct given current vector in the primary winding, the the flux ﬂux linkage linkage produced produced by by the the primaryprimary winding currentcurrent waswas stationarystationary and and constant, constant, and and its its value value could could be be adjusted adjusted by by varying varying the * theipd .ipd The*. The primary primary winding winding current current vector vector was was set stationary set stationary and theand secondary the secondary winding winding current current vector vectorwas oriented was oriente by θrd. Moreover,by θr. Moreover, the maximum the maximum torque torque per ampere per ampere (MTPA) (MTPA) strategy strategy was used was to used control to * * controlthe secondary the secondary winding winding current currentisd and isdi*sq and. For isq*. the For DC the excitationDC excitation control, control, Equation Equation (6) becomes: (6) becomes:

33 TTe “ pr pLps Li iisq i ((18)18) e22 r pspd pd sq

TheThe decoupled control of Te was realized through iisqsq according to to E Equationquation ( (18);18); moreover, moreover, the the torquetorque capacity capacity could could be be increased increased by controlling t thehe DC excitation in the primary winding.

44.1.2..1.2. Simulation Simulation S Studiestudies TheThe simulation studies studies of of the the proposed proposed DC DC excitation excitation control control method method were carriedwere carried out in Matlab out in MatlabSimulink Simulink using the using DSFM the motor DSFM data motor from Table data1 obtained from Table by Maxwell 1 obtained model. by TheMaxwell simulation model. results The simulation results in Figure 9 have been successfully experimentally verified by the corresponding experiments. Both the simulation results and the experimental results will be jointly discussed in the following section.

120 6

110 5 )

m 100 4 p r (

d 90 3 e e p S

80 2 n e

Torque (Nm) Torque v i 70 1 G

60 0

50 -1 0.3 0.4 0.5 0.6 0.7 0.8 0.3 0.4 0.5 0.6 0.7 0.8 Time (s) Time (s) (a) (b)

Energies 2016, 9, 517 8 of 20

Figure 8. DC excitation control method.

With the direct given current vector in the primary winding, the flux linkage produced by the primary winding current was stationary and constant, and its value could be adjusted by varying the ipd*. The primary winding current vector was set stationary and the secondary winding current vector was oriented by θr. Moreover, the maximum torque per ampere (MTPA) strategy was used to control the secondary winding current isd* and isq*. For the DC excitation control, Equation (6) becomes: 3 TpLii= (18) e2 r ps pd sq

The decoupled control of Te was realized through isq according to Equation (18); moreover, the torque capacity could be increased by controlling the DC excitation in the primary winding.

Energies 2016, 9,4.1.2. 517 Simulation Studies 8 of 19 The simulation studies of the proposed DC excitation control method were carried out in Matlab Simulink using the DSFM motor data from Table 1 obtained by Maxwell model. The in Figure9 havesimulation been successfullyresults in Figure experimentally 9 have been successfu veriﬁedlly experimentally by the corresponding verified by the corresponding experiments. Both the simulation resultsexperiments. and theBoth experimentalthe simulation results results and the will expe berimental jointly results discussed will be jointly in the discussed following in the section. following section.

2 Torque (Nm) Torque 1

-1 0.3 0.4 0.5 0.6 0.7 0.8 Time (s) Energies 2016, 9, 517 9 of 20 (a) (b) 120 130 125 110 120

100 115 ) )

m m 110 p p

r 90 r ( ( 105 d d e e

e 80 e 100 p p S S 70 95 90 60 85

50 80 0.3 0.4 0.5 0.6 0.7 0.8 0.3 0.4 0.5 0.6 0.7 0.8 Time (s) Time (s) (c) (d) 8 20

15 6 ) ) A A ( ( 4

10 t t n n e e r

5 r 2 r r u u C C 0

0 y y r r a

-5 a d

d -2 n n o -10 o c c

e -4 e S -15 S -6 -20 -8 0.3 0.4 0.5 0.6 0.7 0.8 0.3 0.4 0.5 0.6 0.7 0.8 Time (s) Time (s) (e) (f) 5 5 4.5 4.5

) 4 ) 4 A A ( (

3.5 3.5 t t n n

e 3 e 3 r r r r

u 2.5 u 2.5 C C

y y

r 2 r 2 a a

m 1.5 m 1.5 i i r r

P 1 P 1

0.5 0.5

0 0 0.3 0.4 0.5 0.6 0.7 0.8 0.3 0.4 0.5 0.6 0.7 0.8 Time (s) Time (s) (g) (h)

Figure 9. Simulation results of the DC excitation control for the DSFM motor: (a) curve of the given Figure 9. Simulationspeed variation; results (b) curve of the of the DC load excitation torque variation; control (c) speed for the curve DSFM for speed motor: varying (a from) curve 55 to of the given speed variation;110 rpm (b; )(d curve) speed ofcurve the for load torque torque varying variation; from 0 to 5 Nm (c); speed (e) secondary curve current for speed curve for varying speed from 55 to 110 rpm; (d)varying speed from curve 55 to for 110 torque rpm; (f) varying secondary from current 0 curve to 5 Nm; for torque (e) secondary varying from current 0 to 5 Nm curve; (g) for speed primary current curve for speed varying from 55 to 110 rpm; and (h) primary current curve for varying fromtorque 55 to varying 110 rpm; from (0f to) secondary5 Nm. current curve for torque varying from 0 to 5 Nm; (g) primary current curve for speed varying from 55 to 110 rpm; and (h) primary current curve for torque varying from 0 to4.1. 53 Nm.. Experimental Results Figure 10a,b shows the actual DSFM motor test rig and schematic of its constituent parts, respectively.

Energies 2016, 9, 517 9 of 19

4.1.3. Experimental Results Figure 10a,b shows the actual DSFM motor test rig and schematic of its constituent parts,Energies respectively. 2016, 9, 517 10 of 20 Energies 2016, 9, 517 10 of 20

Dual-Fed Flux-Modulated Servo motor Servo motor Torque Dual-Fed Fmluoxto-Mr odulated Servo motor DSFM motor Se（rvolo mado）tor Tsoernqsuoer motor DSFM motor （load） sensor Resolver

Signal Resolver U V W U1V1W1 U2V2W2 trSanigsndaulc er DC SUer v oV m Wotor U1V1W1 U2V2W2 U1+ transducer Back to Back Controller U1c+ontroller DPCow er Servo motor converter DC UU11+- U1- +12V Back to Back Controller P(otowrqeur e) U1c+ontroller U+ LDiCnk U1- U1- +1D2CV Uc-onverter Torque sensor (torque) Power U+ Link DC U- Torque sensor DC power supply Power Oscilloscope Multimeter DC power supply Can Bus Oscilloscope Multimeter PC Can Bus PC (a) (b) (a) (b) FigureFigure 10. 10.Experimental Experimental environmentenvironment of thethe DSFM motor:motor: ( aa)) l laboratoryaboratory test test facility facility;; and and (b (b) )b blocklock Figure 10. Experimental environment of the DSFM motor: (a) laboratory test facility; and (b) block diagramdiagram of of the the DSFM DSFM motormotor experimentalexperimental setup.setup. diagram of the DSFM motor experimental setup. The DSFM motor can operate stably with DC excitation in the primary winding. The current curvesTheThe of DSFMDSFM the DSFM motormotor motor cancan operateatoperate 200 rpm stablystably and with3 Nm DC are excitationshown in Figurein in the primary 11. winding winding.. The The current current curvescurves ofof thethe DSFMDSFM motormotor atat 200200 rpmrpm and 3 Nm are shown in Figure 1111.. 25 25 25 25 20

)

) 20 A

) 15 ( 20 A e ) ( A

( t 15 20 A e 10 u ( n t d

i e t l u r 10 n t r p

5 i 15 e l u r m r p

C 5 15

u 0 A m

y C r c 0 A i a

y -5 10 d n r c i a n -5 o 10 d n o -10 m c n o r e o -10 a m c S -15 5 r e H a

S -15 5 -20 H -20 -25 0 0 0.02 0.04 0.06 0.08 0.1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 -25 0 0 0.02 0.0T4 ime (0s.)06 0.08 0.1 0 1 2 3 4 5 H6 ar7mo8nic9 O10rd1e1r 12 13 14 15 Time (s) Harmonic Order (a) (b) (a) (b) Figure 11. Winding currents of the DSFM motor at 200 rpm: (a) secondary winding current; and (b) FigureFigurethe harmonic 11.11. WindingWinding content currentscurrents of the of currentof the the DSFM DSFM. motor motor at at 200 200 rpm: rpm (a: )(a secondary) secondary winding winding current; current and; and (b )(b the) harmonicthe harmonic content content of the of current.the current. As shown in Figure 11a, the secondary current was asymmetric sine wave and was stable at 36.6 Hz whenAs shown the DC in Figure current 11a in, tthehe secondaryprimary winding current was asymmetric8 A and the sine given wave speed and was was 200stable rpm. at 36.6The As shown in Figure 11a, the secondary current was asymmetric sine wave and was stable at Hzharmonic when thecontent DC currentof the curren in thet wasprimary low aswinding shown wasin Figure 8 A and 11b .the given speed was 200 rpm. The 36.6 Hz when the DC current in the primary winding was 8 A and the given speed was 200 rpm. The harmonicTo evaluate content theof the performance current was of low the as proposed shown in control Figure method11b. of the DSFM motor, the speed harmonic content of the current was low as shown in Figure 11b. responseTo evaluate was analysed the performance by changing of the the proposed given speedcontrol or method increasing of the the DSFM load motor, torque, the and speed the To evaluate the performance of the proposed control method of the DSFM motor, the speed responsecorresponding was analysed phase current by changings in both the primary given speed and secondary or increasing windings the load were torque, measured. and The the response was analysed by changing the given speed or increasing the load torque, and the correspondingexperimental results phase are current showns in in bothFigure the 1 2 primary. and secondary windings were measured. The corresponding phase currents in both the primary and secondary windings were measured. The experimental results are shown in Figure 12. experimental results are shown in Figure 12.

Energies 2016, 9, 517 10 of 19 Energies 2016, 9, 517 11 of 20

130 7 120 6 110 ) 5 m )

p 100

r 4 ( m 90 N d (

e 3

e 80 e p u S q 2

n 70 o e

T 1 v

i 60 G 0 50

40 -1

30 -2 0 2 4 6 8 10 12 14 5 10 15 Time (s) Time (s) (a) (b) 120 130 125 110 120

100 115 ) ) m m p p 110 r r 90 ( (

105 d d e e e 80 e 100 p p S S 70 95 90 60 85

50 80 0 2 4 6 8 10 12 14 5 10 15 Time (s) Time (s) (c) (d) 10 10 8 ) )

6 A A (

( t

t 5 4 n n e r e r r r

2 u u C

0 0 y y r r a a

-2 d d n n o o -4 -5 c c e e S S -6

-8 -10

-10 0 2 4 6 8 10 12 14 5 10 15 Time (s) Time (s) (e) (f) 50 50

40 40 ) ) A A

( 30 ( 30

t t n n e e

r 20 r 20 r r u u C C

10 10 y y r r a a 0 0 m m i i r r P P -10 -10

-20 -20

0 2 4 6 8 10 12 14 5 10 15 Time (s) Time (s) (g) (h)

Figure 12. Experimental results of the DC excitation control for the DSFM motor: (a) curve of the given speed variation; (b) curve of the load torque variation; (c) speed curve for speed varying from 55 to 110 rpm; (d) speed curve for torque varying from 0 to 5 Nm; (e) secondary current curve for speed varying from 55 to 110 rpm; (f) secondary current curve for torque varying from 0 to 5 Nm; (g) primary current curve for speed varying from 55 to 110 rpm; and (h) primary current curve for torque varying from 0 to 5 Nm. Energies 2016, 9, 517 11 of 19

According to Figure 12a, the given speed was increased from 55 to 110 rpm at 1 s; the ﬁnal speed was stable at 110 rpm. In Figure 12c, the rotor speed was changed by a slow progressive change of the PI control. The secondary winding current was changed during the adjustment period, and the stability was restored by increasing the frequency only twice from 10 to 20 Hz, as shown in Figure 12e. The primary winding current was maintained at a constant value, as shown in Figure 12g, because a DC current was supplied to the primary winding all the time. The steady state errors of the speed were less than 5 rpm. By maintaining the given speed at 109 rpm and increasing the load torque from 0 to 5 Nm at 5 s shown in Figure 12b, the speed of the DSFM motor changed to 109 rpm accurately through a period of adjustment. The speed curve is presented in Figure 12d. The plot in Figure 12f illustrates that the amplitude of the secondary winding current was increased with this load torque, and its frequency was maintained at 20 Hz. The increase in the secondary winding current resulted from the increase in isq, thus enhancing the electromagnetic torque of the DSFM motor Te as predicted by Equation (18). The current in the primary winding was maintained at a constant value, as shown in Figure 12h. The measured and simulated results in Figure9 are mostly identical to the experimental. The simulated speed responses were better than the experimental because the PI parameters for speed control experiment were not ideal.

4.1.4. Discussion According to the preceding simulation and experiment results, the proposed DC excitation control can realize the decoupled control for the current of the two stator windings and the electromagnetic torque Te. For this control method, the DSFM motor can always operate under motoring conditions similar to PMSMs, which allows the application of the DSFM motor to EVs.

4.2. SCOVC with AC Excitation

4.2.1. Theoretical Analysis

If ωp in Equation (11) is not equal to 0 then, according to Equation (4), the conversion formulas of the primary frame become: υ υ ´jθp pr “ ps e ´jθp (19) ipr “ ips e and the conversion formulas of the secondary become:

υ “ υ e´jθs sr ss (20) ´jθs isr “ iss e

This relationship must be satisﬁed at any time so that the secondary frame can be oriented according to the rotor angle θr and angle of the primary winding current θp. The SCOVC with AC excitation is referred to as AC excitation control in the following text. The AC excitation control method for the DSFM motor is illustrated in Figure 13. The current closed loop control with a given fp was implemented to produce a rotating magnetic ﬁeld with angular velocity ωp, and the primary winding current phase angle θp was established after the integral link. The secondary winding current phase angle θs was then obtained according to Equation (13). Thus, the secondary winding current vector is was oriented to the ds axis by θs. The two winding current vectors were oriented throughout the AC excitation control method. Moreover, the frequency of the secondary winding changed to fs with a known θr, which implied that the speed control of the DSFM motor was indirectly conducted using the frequency of the secondary winding * current. For example, if ωr = 218 rpm and fp = 20 Hz, fs = 20 Hz can be exactly acquired using the algorithm of the proposed control method, and the rotor can be operated accurately at 218 rpm. Energies 2016, 9, x 13 of 14 Energies 2016, 9, 517 13 of 20

−θ υ=υe j p pprsi iA p−θα (19) iie= j p iB Papprsrk ipβ Clarke iC and the conversion formulasi of the secondary become: i* pd upd pd PI upα−θ υ=υ j s * i u InvPassrk upβe SVPWM IGBT i =0 pq PI pq rs pq −θ (20) = j s 2π ∫ iiess outer Energies 2016, 9, 517 fp θ rs 12 of 19 ωp p Current Vector stator Oriented Control rotor This relationship* must be satisfiedusd at any time so that the secondary frame can be oriented i sd=0 PI usα according to the rotor angle θr and angle of the primary winding current θp. The SCOVCinner with AC isd InvPark usβ SVPWM IGBT Moreover,excitation the MTPA is* referred control to wasas AC used excitationusq to realize control thein the decoupled following text. control The AC of Texcitationset,a whichtor control was expressed ω r PI method for the DSFM motor is illustrated in Figure 13. using Equation (18). ωr isq ia outer isα Park ib stator isβ Clarke ic

θr K Figure 13. AC excitation control method.

The current closed loop control with a given fp was implemented to produce a rotating magnetic field with angular velocity ωp, and the primary winding current phase angle θp was established after the integral link. The secondary winding current phase angle θs was then obtained according to Equation (13). Thus, the secondary winding current vector is was oriented to the ds axis by θs. The two winding current vectors were oriented throughout the AC excitation control method. Moreover, the frequency of the secondary winding changed to fs with a known θr, which implied that the speed control of the DSFM motor was indirectly conducted using the frequency of the secondary winding current. For example, if ωr* = 218 rpm and fp = 20 Hz, fs = 20 Hz can be exactly acquired using the algorithm of the proposed control method, and the rotor can be operated accurately at 218 rpm. Moreover, the MTPA control was used to realize the decoupled control of Te, which was expressed Figure 13. AC excitation control method. using Equation (18). Figure 13. AC excitation control method.

The current closed loop control with a given fp was implemented to produce a rotating magnetic field with 4.2.2.4.2. Simulation2. Simulationangular velocity Results Results ωp, and the primary winding current phase angle θp was established after the integral link. The secondary winding current phase angle θs was then obtained according to Equation (13). Thus, the secondary ToTo fully fully test test the the performance performance of of the the proposedproposed control strategy, start start-up-up of of the the motor motor under under load load winding current vector is was oriented to the ds axis by θs. The two winding current vectors were oriented waswas simulated, simulated,throughout and and the the AC the excitation results results control are are shown shown method. in inMoreover, FigureFigure the 1414. frequency. The given given of the secondaryspeed speed was waswinding 500 500 changedrpm rpm and andto f s the the load load torquetorque was waswith 1.4 a1.4 known Nm. Nm. θ r, which implied that the speed control of the DSFM motor was indirectly conducted using the frequency of the secondary winding current. For example, if ωr* = 218 rpm and fp = 20 Hz, fs = 20 Hz can be 600 exactly acquired using the algorithm of the proposed control method,600 and the rotor can be operated accurately at 218 rpm. Moreover, the MTPA control was used to realize the decoupled control of Te, which was expressed 500 using Equation (18). 500

400 400 ) ) 4.2.2. Simulation Results m m p p

r 300

r 300 To fully test the performance of the proposed control ( strategy, start-up of the motor under load was (

d d

simulated, and the results are shown in Figure 14. The given speede was 500 rpm and the load torque was 1.4 Nm. e

e 200

e 200 p p S S 100 100

0 0

-100 -100 0 0.1 0.2 0.3 0.4 0.5 0.6 0 0.2 0.4 0.6 0.8 1 1.2 Time (s) Time (s) Energies 2016, 9, 517 14 of 20 (a) (b) 2.5 7

6 2 5 ) ) m m 1.5 N N 4 ( (

e e u u q q 3 r r 1 o o T T 2 0.5 1

0 0 0 0.2 0.4 0.6 0.8 1 1.2 0 0.1 0.2 0.3 0.4 0.5 0.6 Time (s) Time (s) (c) (d) 25 25 20 Figure 14. Cont20. ) ) 15 15 A A (

(

t t 10 10 n n e e r r r r 5 5 u u C C

0 0 y y r r a a -5 -5 d d n n

o -10

o -10 c c e e -15 -15 S S

-20 -20

-25 -25 0 0.2 0.4 0.6 0.8 1 1.2 0 0.1 0.2 0.3 0.4 0.5 0.6 Time (s) Time (s) (e) (f) 6 20 15 4 )

A 10 (

2 t n

e 5 r r 0 u 0 C

y r

a -5 -2 m i r -10

Primary Current (A) Current Primary P -4 -15

-6 -20 0 0.2 0.4 0.6 0.8 1 1.2 0 0.1 0.2 0.3 0.4 0.5 0.6 Time (s) Time (s) (g) (h)

Figure 14. Simulation results of the start-up of the motor under load: (a) speed curve for i*pd = 5 A; (b)

speed curve for i*pd = 15 A; (c) torque curve for i*pd = 5 A; (d) torque curve for i*pd = 15 A; (e) secondary current curve for i*pd = 5 A; (f) secondary current curve for i*pd = 15 A; (g) primary current curve for

i*pd = 5 A; and(h) primary current curve for i*pd = 15 A.

Note from Figure 14 that when the primary current i*pd increased, the starting torque increased and the starting time decreased. The start-up simulation illustrates that Te can be controlled not only by the secondary current i*sq, but also by the primary current i*pd, one can choose the better control method according to the actual situation. More simulation studies of the proposed AC excitation control method were carried out and experimentally verified by the corresponding experiments. The simulation results are shown in Figure 15, and agree well with the experimental results. The corresponding plots will be jointly discussed in the next section.

Energies 2016, 9, 517 14 of 20

2.5 7

6 2 5 ) ) m m 1.5 N N 4 ( (

e e u u q q 3 r r 1 o o T T 2 0.5 1

0 0 Energies02016, 9,0. 5172 0.4 0.6 0.8 1 1.2 0 0.1 0.2 0.3 0.4 0.5 13 of0.6 19 Time (s) Time (s) (c) (d) 25 25 20 20 ) ) 15 15 A A (

(

t t 10 10 n n e e r r r r 5 5 u u C C

0 0 y y r r a a -5 -5 d d n n

o -10

o -10 c c e e -15 -15 S S

-20 -20

-25 -25 0 0.2 0.4 0.6 0.8 1 1.2 0 0.1 0.2 0.3 0.4 0.5 0.6 Time (s) Time (s) (e) (f) 6 20 15 4 )

A 10 (

2 t n

e 5 r r 0 u 0 C

y r

a -5 -2 m i r -10

Primary Current (A) Current Primary P -4 -15

-6 -20 0 0.2 0.4 0.6 0.8 1 1.2 0 0.1 0.2 0.3 0.4 0.5 0.6 Time (s) Time (s) (g) (h)

*pd * Figure 1 14.4. SimulationSimulation results results of of the the start start-up-up of ofthe the motor motor under under load load:: (a) s (apeed) speed curve curve for i for = i5pd A;= (b 5) *pd * *pd * *pd * A;speed (b) curve speed for curve i = for 15 iApd; (c=) 15torque A; ( ccurve) torque for curvei = 5 forA; (idpd) torque= 5 A; curve (d) torque for i curve = 15 A for; (e)i secondarypd = 15 A; * * * * current(e) secondary curve for current i pd = curve5 A; (f for) secondaryi pd = 5 A; current (f) secondary curve for current i pd = 15 curve A; (g for) primaryi pd = 15current A; (g )curve primary for * * * * icurrentpd = 5 A curve; and(h for) pirimarypd = 5 A;current and(h curve) primary for i currentpd = 15 A curve. for i pd = 15 A.

* Note from Figure 14 that when the primary current i* pd increased, the starting torque increased Note from Figure 14 that when the primary current i pd increased, the starting torque increased and the starting time decreased. The start-up simulation illustrates that Te can be controlled not and the starting time decreased. The start-up simulation illustrates that Te can be controlled not only * * only by the secondary current* i sq, but also by the primary current* i pd, one can choose the better by the secondary current i sq, but also by the primary current i pd, one can choose the better control control method according to the actual situation. method according to the actual situation. More simulation studies of the proposed AC excitation control method were carried out and More simulation studies of the proposed AC excitation control method were carried out and experimentally verified by the corresponding experiments. The simulation results are shown in experimentally veriﬁed by the corresponding experiments. The simulation results are shown in Figure 15, and agree well with the experimental results. The corresponding plots will be jointly Figure 15, and agree well with the experimental results. The corresponding plots will be jointly discussed in the next section. discussed in the next section.

Energies 2016, 9, 517 14 of 19 Energies 2016, 9, 517 15 of 20

6 220 5

) 200

m 4 p r (

180 d

e 3 e

p 160 S 2 n e v

140 (Nm) Torque i 1 G

120 0

100 -1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.3 0.4 0.5 0.6 0.7 0.8 Time (s) Time (s) (a) (b) 240 240 235 220 230

200 225 ) )

m m 220 p p

r 180 r ( ( 215 d d e e

e 160 e 210 p p S S 140 205 200 120 195

100 190 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.3 0.4 0.5 0.6 0.7 0.8 Time (s) Time (s) (c) (d) 15 20

15 )

) 10 A A ( (

t 10 t n n

e 5 e r 5 r r r u u C C

0 0 y y r r a -5 a d d n n -5 o -10 o c c e e S -15 S -10

-20 -15 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.3 0.4 0.5 0.6 0.7 0.8 Time (s) Time (s) (e) (f) 8 4 6 3 ) )

A 4 A 2 ( (

t t n n

e 2 1 e r r r r u 0 u 0 C C

y y r r a

-2 a -1 m m i i r -4 r -2 P P

-6 -3

-8 -4 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.3 0.4 0.5 0.6 0.7 0.8 Time (s) Time (s) (g) (h)

Figure 15. Simulation results of the AC excitation control for the DSFM motor: (a) curve of the given Figure 15. Simulation results of the AC excitation control for the DSFM motor: (a) curve of the given speed variation. (b) curve of the load torque variation; (c) speed curve for speed varying from 110 to speed variation. (b) curve of the load torque variation; (c) speed curve for speed varying from 110 to

220 rpm; (d) speed curve for torque varying from 0 to 5 Nm; (e) secondary current curve for speed varying from 110 to 220 rpm; (f) secondary current curve for torque varying from 0 to 5 Nm; (g) primary current curve for speed varying from 110 to 220 rpm; and (h) primary current curve for torque varying from 0 to 5 Nm. Energies 2016, 9, 517 16 of 20

220 rpm; (d) speed curve for torque varying from 0 to 5 Nm; (e) secondary current curve for speed varying from 110 to 220 rpm; (f) secondary current curve for torque varying from 0 to 5 Nm; (g) Energiesprimary2016, 9 ,current 517 curve for speed varying from 110 to 220 rpm; and (h) primary current curve for15 of 19 torque varying from 0 to 5 Nm.

44.2.3..2.3. Experimental Experimental R Resultsesults The DSFM motor can operate at dif differentferent speeds with different load torques. The current current curves of the DSFM motor at 924 rpm and 3 Nm are shown in Figure 1616..

15 20 15

10 ) ) A 10 ( A

( t

n t 5 5 e n r e r r u r 0 u C

0 C y

r -5 y a r d a

-5 n -10 m o i r c e P -15 S -10 -20

-15 -25 -0.1 -0.05 0 0.05 0.1 -0.1 -0.05 0 0.05 0.1 Time (s) Time (s) (a) (b) 14 20 18 ) ) 12 A A ( ( 16

e e d d 10 14 u u t t i i l l 12 p p 8 m m 10 A A

c 6 i i 8 n n o o 6

4 m m r r a a 4 H H 2 2

0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Harmonic Order Harmonic Order (c) (d)

Figure 16.16. WindingWinding currents currents of theof the DSFM DSFM motor motor at 924 at rpm: 924 ( a rpm) primary: (a) p windingrimary windingcurrent; ( b current) secondary; (b) secondarywinding current; winding (c) current the harmonic; (c) the content harmonic of the content primary of current;the primary and (current;d) the harmonic and (d) contentthe harmonic of the contentsecondary of the current. secondary current.

As shown in Figure 16a,b, the two winding currents are asymmetric sine waves and are stable As shown in Figure 16a,b, the two winding currents are asymmetric sine waves and are stable at 85 Hz when the given speed was 924 rpm. The harmonic content of the currents were both low as at 85 Hz when the given speed was 924 rpm. The harmonic content of the currents were both low as shown in Figure 16c,d. shown in Figure 16c,d. Then the speed responses and phase currents are illustrated in Figure 17. Then the speed responses and phase currents are illustrated in Figure 17.

Energies 2016, 9, 517 16 of 19 Energies 2016, 9, 517 17 of 20

240 6

220 5 ) 200 m 4 ) p r ( m

180 N d

( 3 e

e 160 e p u S q

2 r n 140 o e T v i 1

G 120

100 0

80 -1 0 2 4 6 8 10 12 14 5 10 15 Time (s) Time (s) (a) (b) 240 240 235 220 230

200 225 ) ) m m

p 220 p 180 r r ( (

d 215 d e e e e 160 210 p p S S 205 140 200 120 195

190 100 5 10 15 0 2 4 6 8 10 12 14 Time (s) Time (s) (c) (d) 10 20 8 15 ) ) 6 A A (

( 10 4

5 t

2 n n e e r r r r u

u 0 0 C C

y y r r -2 a a -5 d d n n o -4 o c c e e -10 S -6 S -15 -8

-10 -20 0 2 4 6 8 10 12 14 5 6 7 8 9 10 11 12 13 14 15 Time (s) Time (s) (e) (f) 10 10 8

5 ) 6 ) A A ( ( 4

t 0 2 t n n e e r r r r 0 u u C C

-5 y -2 r r a a m i m

r -4 i r P P -10 -6

-8

-15 -10 0 2 4 6 8 10 12 14 5 6 7 8 9 10 11 12 13 14 15 Time (s) Time (s) (g) (h)

Figure 17. Experimental results of the AC excitation control for the DSFM motor: (a) curve of the

given speed variation. (b) curve of the load torque variation; (c) speed curve for speed varying from 110 to 220 rpm; (d) speed curve for torque varying from 0 to 5 Nm; (e) secondary current curve for speed varying from 110 to 220 rpm; (f) secondary current curve for torque varying from 0 to 5 Nm; (g) primary current curve for speed varying from 110 to 220 rpm; and (h) primary current curve for torque varying from 0 to 5 Nm. Energies 2016, 9, 517 17 of 19

There are two processes in the AC excitation control experiments as shown in Figure 17a. First, the given speed was increased from 110 to 220 rpm at 3 s. Second, the frequency of primary winding current was increased from 10 to 20 Hz at 8 s. According to Figure 17c, the ﬁnal speed was stable at 220 rpm throughout a slow progressive change of the PI control. The secondary winding current changed during the two periods of adjustment. First, from 3–8 s, its frequency increased from 10 to 30 Hz. Second, from 8–10 s, its frequency decreased from 30 to 20 Hz, and was ﬁnally stable after 14 s. The secondary current curve is shown in Figure 17e. The amplitude of the primary winding current restored stability and the frequency was changed to 20 Hz through an adjustment period of 2 s from 8–10 s as shown in Figure 17g. The steady state errors of the speed were less than 5 rpm. By maintaining the given speed at 218 rpm and increasing the load torque from 0 to 5 Nm at 5 s shown in Figure 17b, the speed of the DSFM motor changed to 218 rpm accurately through a period of adjustment. The speed curve is presented in Figure 17d. The plot in Figure 17f illustrates that the amplitude of the secondary winding current was increased using this load torque, and its frequency was maintained at 20 Hz. The current in the primary winding was controlled using the current closed loop control presented in Figure 11 and maintained constant, as shown in Figure 17h. The same speed, frequency and load torque change was implemented as shown in Figure 15, and the results are also the same. The speed can be strictly controlled by the speed closed loop control, and the frequency of the two winding currents can be automatically adjusted through the AC excitation control method.

4.2.4. Discussion The experimental results showed high speed response with the change in speed and load torque, which illustrated that the DSFM motor can operate stably and reliably under different conditions. The wide speed range of the DSFM motor can be achieved without flux-weakening control such that the copper loss can be reduced and the efficiency can be increased at a high speed compared with PMSMs. Thus, the proposed control can meet the requirement of high-speed operation and efficient torque control, and would have great application prospects in EVs.

5. Conclusions This study explored and investigated the control strategies for the application of the DSFM motor to EVs. Two available vector control methods were proposed and control system effects were validated using a customised DSFM prototype.

1. The differences in air gap structures of the DSFM and BDFM motors were compared and analysed, and the mathematical model of the DSFM motor was used as the basis of the decoupled control method. 2. The VVVF control was implemented to explore and verify the operating principles of the DSFM motor. According to the experimental results of the VVVF control, the speed and torque of the DSFM motor can be controlled using the frequency of the current in the two stator windings. 3. The SCOVC with DC excitation control method was realized theoretically and experimentally, and high speed and torque performance were obtained through the variable speed and torque tests. 4. The SCOVC with AC excitation control method was realized theoretically and experimentally, and high speed and torque performance were obtained through the variable speed and torque tests.

The DSFM motor and its control strategies proposed in this paper can serve as a basis for further research on application of the ﬂux-modulated motor to EVs, and can be crucial for more applications of such motors. The DC excitation control had a limited range of frequency, while the AC one did not. Thus, the two closed loop controls can be ﬂexibly applied to different situations. For example, in generator mode, the DSFM motor can be used as a DC generator with the DC one or an AC generator with the AC one, and in motor mode, it can be used as a slow speed motor with the DC one or a high speed motor with Energies 2016, 9, 517 18 of 19 the AC one. The proposed control methods will be implemented on an electric bus being developed in our laboratory.

Acknowledgments: This study was supported in part by the National Science Foundation of China (51477058), Science and Technology Plan Project of Xiamen (3502Z20153029), and Promotion Program for Young and Middle-aged Teacher in Science and Technology Research of Huaqiao University (ZNQ-YX304). Author Contributions: Xinhua Guo and Weinong Fu conceived and designed the experiments; Shaozhe Wu, Yunchong Wang, and Peihuang Zeng performed the experiments; Shaozhe Wu performed the simulations; Shaozhe Wu and Yulong Liu analysed the data; and Xinhua Guo wrote the paper. Conﬂicts of Interest: The authors declare no conﬂict of interest.

Abbreviations The following abbreviations are used in this paper: DSFM Dual-stator ﬂux-modulated BDFMs Brushless doubly-fed machines BDFG Brushless doubly-fed generator BDFRM Brushless doubly-fed reluctance machine EV Electric vehicle SCOVC Stator-current-oriented vector control DC Direct current AC Alternating current VC Voltage vector-oriented control FOC Field-oriented control PMSMs Permanent magnet synchronous motors VVVF Variable voltage and variable frequency MTPA Maximum torque per ampere

References

1. Long, B. Energy-regenerative braking control of electric vehicles using three-phase brushless direct-current motors. Energies 2014, 7, 99–114. [CrossRef] 2. Cheng, M.; Sun, L.; Buja, G.; Song, L.H. Advanced electrical machines and machine-based systems for electric and hybrid vehicles. Energies 2015, 8, 9541–9564. [CrossRef] 3. Chan, C.C. The state of the art of electric, hybrid, and fuel cell vehicles. Proc. IEEE 2007, 95, 704–718. [CrossRef] 4. Chan, C.C.; Wong, Y.S. The state of the art of electric vehicles technology. In Proceedings of the 4th International Power Electronics and Motion Control Conference, Xi’an, China, 14–16 August 2004; pp. 46–57. 5. Wang, Y.C.; Ho, S.L.; Fu, W.N.; Shen, J.X. A novel brushless doubly fed generator for wind power generation. IEEE Trans. Magn. 2012, 48, 4172–4175. [CrossRef] 6. Knight, A.; Betz, R.; Dorrell, D.G. Design and analysis of brushless doubly fed reluctance machines. IEEE Trans. Ind. Appl. 2013, 49, 50–58. [CrossRef] 7. Hsieh, M.F.; Lin, I.H.; Dorrell, D.G. Magnetic circuit modeling of brushless doubly-fed machines with induction and reluctance rotors. IEEE Trans. Magn. 2013, 49, 2359–2362. [CrossRef] 8. Roberts, P.; Long, T.; McMahon, R.; Shao, S.; Abdi, E.; Maciejowski, J. Dynamic modelling of the brushless doubly fed machine. IET Elect. Power Appl. 2013, 7, 544–556. [CrossRef] 9. Knight, A.M.; Betz, R.E.; Song, W.K.; Dorrell, D.G. Brushless doubly-fed reluctance machine rotor design. In Proceedings of the IEEE Energy Conversion Congress and Exposition (ECCE), Raleigh, NC, USA, 15–20 September 2012; pp. 2308–2315. 10. Logan, T.; McMahon, R.; Seffen, K. Noise and vibration in brushless doubly fed machine and brushless doubly fed reluctance machine. IET Elect. Power Appl. 2013, 8, 50–59. [CrossRef] 11. Jovanovic, M. Sensored and sensorless speed control methods for brushless doubly fed reluctance motors. IET Elect. Power Appl. 2009, 3, 503–513. [CrossRef] 12. Ademi, S.; Jovanovic, M. Vector control strategies for brushless doubly-fed reluctance wind generators. In Proceedings of the 2nd International Symposium on Environment-Friendly Energies and Applications (EFEA), Newcastle upon Tyne, UK, 25–27 June 2012; pp. 44–49. Energies 2016, 9, 517 19 of 19

13. Ademi, S.; Jovanovic, M. Theoretical and experimental evaluation of vector control for doubly-fed reluctance generators. In Proceedings of the International Conference on Electrical Machines (ICEM), Berlin, Germany, 2–5 September 2014; pp. 936–942. 14. Ademi, S.; Jovanovic, M. Maximum torque per inverter ampere control of brushless doubly-fed reluctance generators for wind turbines. In Proceedings of the International Symposium on Power Electronics, Electrical Drives, Automation and Motion (SPEEDAM), Ischia, Italy, 18–20 June 2014; pp. 883–888. 15. Ademi, S.; Jovanovic, M.; Hasan, M. Control of brushless doubly-fed reluctance generators for wind energy conversion systems. IEEE Trans. Energy Convers. 2015, 30, 596–604. [CrossRef] 16. Neti, P.; Nandi, S. Determination of effective air-gap length of synchronous reluctance motors (SynchRel) from experimental data. IEEE Trans. Ind. Appl. 2006, 4, 454–464. [CrossRef] 17. Betz, R.E.; Jovanovic, M.G. Introduction to the space vector modeling of the brushless doubly fed reluctance machine. Elect. Power Compnt. Syst. 2003, 31, 729–755. [CrossRef]

© 2016 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/).