Deployment Framework for the Internet of Water Meters Using Computer Vision on ARM Platform
Total Page:16
File Type:pdf, Size:1020Kb
Journal of Ambient Intelligence and Smart Environments 12 (2020) 35–60 35 DOI 10.3233/AIS-200544 IOS Press Deployment framework for the Internet of water meters using computer vision on ARM platform Gaubert V. Santiago and Alberto J. Alvares * Mechanic and Mechatronic Engineering Department, University of Brasilia, Brasilia-DF, Brazil E-mails: [email protected], [email protected] Abstract. This article presents the conception of a new method developed mainly in Python to automate the reading process of water meters with an analog display using computer vision and machine learning. A camera captures the consumption value in the water meter, and the yielded image undergoes image processing until the digits are detected and isolated. Then the digits are passed into an SVM machine-learning model that carries out a high accuracy OCR. The software is executed over an ARM platform running Linux. The data resultant from the automated metering, such as the device identification number, event date and time in UTC, consumption value, volume and time variations, flow, and display image, are locally stored and transmitted to a cloud server through VPN in a Wi-Fi and cellular network connection, or by SMS, enabling a remote supervision. Thereby, the automatic metering method features a new way to perform predictive analysis and management of water and meters proactively and can be replicated for digital-display water meters, as well as extended to handle automatic metering on electricity and gas meters as well. Keywords: ARM platform, computer vision, internet of things, machine learning, water meter 1. Introduction curate and reliable data for the relevant applica- tion system, or effectively monitor user behavior. With the development of Smart City [15,33], Indus- – High bills can create disputes with customers, try 4.0 [3], and the improvement of the standard of liv- leading to delays in the operation and manage- ing, there has been a rapid augmentation in the num- ment of payment processing. ber of residential customers of gas, water or electric- The objective of the present article is to solve the ity services. As a result, suppliers’ metering fees are above issues by providing a convenient and innova- increasingly prominent. The drawbacks of the billing tive method of metering automation that yields accu- and payment management of manual metering become rate data collected from the automatic metering of me- an obstacle for metering companies to improve effi- ter instruments by using computer vision and the Inter- ciency, management, and service levels. Below are the net of Things (IoT). problems in the conventional, manual metering: With the advent, popularization, and falling costs of – Access to the house is difficult because of security ARM (Advanced RISC Machine) computer-based ma- × × and privacy. chines with the size of a credit card (85 56 17 mm) – The data collection is not timely, and cannot re- and sufficient processing power to run computer vision flect the real-time system status, nor provide ac- applications, the use of boards such as Raspberry Pi, Lemon Pi, Banana Pi, and Orange Pi becomes cost- effective. With less than 35 dollars, it is possible to *Corresponding author. E-mail: [email protected]. have an HD camera, Wi-fi, Ethernet, USB, cellular net- 1876-1364/20/$35.00 © 2020 – IOS Press and the authors. All rights reserved 36 G.V. Santiago and A.J. Alvares / Framework for the Internet of water meters using computer vision on ARM work, GPIO pins, among other features. This set of de- Sun et al. [38] make some systematical reviews of vice features allows the process of automatic reading the development and deployment of smart energy me- of flow-instruments by supporting and handling the ex- ters, including smart electricity meters, smart heat me- ecution of image processing and machine learning de- ters, and smart gas meters. By examining various func- ployment, as well as data transmission through the In- tions and applications of smart energy meters, as well ternet or SMS. as associated benefits and costs, their investigation pro- This work presents a new method developed in vides insights and guidelines regarding the future de- Python that uses OpenCV (Open Source Computer Vi- velopment of smart meters. Before smart meters, con- sion Library) to automate the reading process of instru- ventional electromechanical meters were the primary ments such as water, gas, and electricity meters with type of devices for measuring electricity flow. These analog or digital display. Briefly, the pipelined algo- old meters usually display the consumption value on rithm of the method captures the metering value and an analog counter, which requires personnel reading to processes its image to identify the digits shown on the assess usage. Smart meters are electronic devices that meter display. It uses machine learning to predict the measure consumption and operate in two-way commu- consumption value, and that data digitalized from the nication regarding the usage information and billing, metering is published to enable asset management and besides providing network status of the grid. predictive analysis. Thus, the method adheres to the As claimed by Sun et al. [39], the significant in- Internet of Things [2] applied to the automatic reading crease in energy consumption and the rapid develop- of meters. ment of renewable energy, such as solar power and This article is organized as follows: Section II wind power, have brought considerable challenges to presents related work; Section III introduces an ar- energy security and the environment, which, in the chitecture proposal for analog metering digitalization meantime, stimulate the development of energy net- through computer vision; Section IV goes over the works toward a more intelligent direction. Smart me- computer-vision method that predicts digits of con- ters are the most fundamental components in intelli- sumption values taken from a water meter display; gent energy networks (IENs). In addition to measur- Section V explains the software implementation; Sec- ing energy flows, smart energy meters can exchange information on energy consumption and the status of tion VI focus on the experimental setup, performance energy networks between utility companies and con- evaluation, and results; and Section VII concludes this sumers. Furthermore, smart energy meters also allow study and points out future works. monitoring and control of home appliances and other devices according to each costumer’s individual needs. According to Barbierato et al. [7], a new generation 2. Related work smart meters are a crucial enabler of advanced meter- ing infrastructure (AMI) and foster new energy-related Automatic meter reading (AMR) [15,47] is the tech- services such as demand response (DR). Aguirre et al. nology of automatically collecting consumption, diag- [1] and LeMay et al. [22] presented two smart-meter nostic, and status data from water meter or electricity systems that allow bidirectional communication with metering devices and transferring that data to a central a centralized DR management platform. Aguirre et al. database for billing, troubleshooting, and analyzing. [1] presented a new generation smart-meter designed AMR has numerous benefits over manual reading, and to support new requirements for operation and con- some of the most important benefits include: (1) ac- trol of the distribution network grid. LeMay et al. [22] curate meter reading; (2) energy management through described a meter gateway architecture for integrated data graphs; (3) low cost; (4) reliable data transmis- control of loads by energy aggregators. Not only smart sion; (5) improved billing; (6) security for premises; meters are essential in such a context. For instance, (7) less financial burden correcting mistakes; (8) less Mashima and Chen [24] presented a DR system frame- time to obtain meter readings. This technology mainly work leveraging on a DR client mobile app able to di- saves utility providers the expense of periodic trips to rectly control IoT devices according to user policies. each physical location to read a meter. Another advan- In this scenario, AMI [42] and IoT devices (i.e., smart tage is that billing can be based on near real-time con- meters [40] and smart appliances), are key technolo- sumption rather than on estimates based on past or pre- gies to foster novel services in the smart grid [30], dicted consumption. such as the DR in residential contexts. In these regards, G.V. Santiago and A.J. Alvares / Framework for the Internet of water meters using computer vision on ARM 37 smart devices are part of AMI and allow a fine-grained sor, battery, GPRS module, and integrated circuitry. collection of energy measurements. In particular, the To put the work described in this patent into practice, new smart meters [40] can sample data spanning in it requires replacing the home’s current water meter the range of 1 second to 15 minutes [6], depending with a new one built with the mentioned components. on the services to provide. Disaggregating and post- This water meter swapping would undoubtedly turn processing these measurements allow retrieving infor- the solution more costly, both in equipment dimen- mation about consumer behavior in households, such sions and implantation cost. Moreover, the patent can- as appliance activation and energy usage [25]. These not be reusable in other sorts of applications, such as post-processed data can feed other novel energy ser- in the reading of electricity usage. vices, such as DR in households [29]. Metretek Inc. [27,28] presents apparatus and method In the commercial solution [34], there are two al- for remote sensor monitoring, metering and control ternatives to achieve water-usage reading remotely, (U.S. Patent 4,241,237 and U.S. Patent 4,455,453). which is also true for gas meter and electricity me- Their invention relates to remote monitoring systems ter. The first alternative consists of a walk-by method, and, in particular, to an automatic meter reading and which is encouraged when the amount of homes to load management system.