Forecasting of EV Arrivals at Battery Swapping Station using GA-BPNN

Abstract

Electric Vehicles (EV) are gaining popularity from the transportation sector, as it causes less harm to the environment. The battery inside the EV can be refilled using battery charging or battery swapping. As battery swapping method is found to be advantageous over battery charging, Battery Swapping Stations (BSS) is presently the hot topic of research. Forecasting of EV arrivals helps in optimal planning of BSS. Back Propagation Neural Network (BPNN) is frequently used in forecasting. BPNN trained with traditional algorithms such as Levenberg Marquardt (LM) gets stuck at the local optima. This problem can be overcomed using metaheuristic algorithms such as Genetic Algorithm (GA). Thus, in this present work a comparative study on forecasting the EV arrivals at BSS is carried out using LM-BPNN and GA-BPNN. The two models have been simulated using MATLAB/Simulink environment and their performance is analysed using metrics such as Mean Square Error (MSE) and simulation time.