Waterwheel Plant Algorithm and Capsule Attention Convolutional Neural Networks for Optimal Sizing Framework for Photovoltaic-Battery EV Charging Microgrids

The increasing use of electric vehicles (EVs) highlights the importance of energy management (EM) and particularly photovoltaic (PV)-battery microgrids (MGs). However, the conventional optimization methodologies are not always capable of striking an optimal balance between cost, energy, and size of the system, considering uncertainties such as the variability of solar resource and the fluctuating demand of charging of EVs. This paper proposes a hybrid method for the optimal sizing framework for PV-battery EV charging MGs. The proposed method is the combined execution of the waterwheel plant algorithm (WWPA) and capsule attention convolutional neural networks (CACNN). Thus, the proposed method is referred to as the WWPA-CACNN approach. The goal of this work is to achieve optimal sizing of PV-battery systems, enhancing energy utilization, cost-efficiency, and grid independence. The WWPA is used to optimize the sizing of PV panels and battery storage to minimize costs and maximize energy utilization in EV charging MGs. The CACNN is used to predict energy generation, storage, and demand, ensuring accurate forecasting and system adaptability. By then, the proposed method is simulated on the MATLAB platform and compared with various existing methods like particle swarm optimization (PSO), artificial neural network (ANN), non-dominated sorting genetic algorithm-II (NSGA-II), modified snake optimization (MSO), and dung beetle optimizer (DBO). The proposed WWPA-CACNN method also has the lowest total lifetime cost of $12 730 and high efficiency of 96%, which underlines its better overall performance to effectively manage PV-battery EV charging MGs at optimal cost.