PARTICLE SWARM OPTIMIZATION AND DIFFERENTIAL EVOLUTION ALGORITHM FOR ENERGY MANAGEMENT

Abstract

The transition of conventional electric grid to smart grid brings with it many concerns and challenges. One such challenge is efficient and smart usage of energy. Energy Management System (EMS) combining both hardware and software is intended at minimizing energy wastage, reduce cost and thus provide eco-friendly solutions. It is a collection of computerized tools used to monitor, control, and optimize the system performance. Various aspects of energy management are home energy management system, building energy management system, advanced metering infrastructure, electric vehicle and demand side energy management system. Home energy management system, building energy management system or grid energy management system acts as interface between energy suppliers and consumers. Efficient use of energy at grid level and end-user level is achieved with advanced metering infrastructure. Initial development of EMS could not incorporate the introduction of Electric Vehicles; however with the design of multi-level EMS it is made possible. EMS together with Internet of Things, Machine Learning/Artificial Neural Network can provide best solutions for economic development. Use of several stochastic and deterministic algorithms for the purpose of energy management has rapidly gained substantial importance recently owing to their flexibility, ease of implementation and efficient results. The present work makes use of two algorithms based on Particle Swarm Optimization (PSO) and Differential Evolution (DE) algorithm for optimizing consumption of power using smart energy meter data analytics. Both are meta-heuristic algorithms that can easily solve multifaceted problems in engineering. The power consumption equation has been optimized using first PSO then DE. The validity and efficiency of the algorithm is tested using the real time data obtained from smart energy meter and a comparative study is presented. The modeling and simulation is carried on MATLAB platform and the results depict that both presented algorithms can ominously reduce the power consumption. With PSO approximately 11.5% reduction in power can be obtained and DE can reduce up to 9.4% power in best-case scenario making PSO superior to DE.