Sand Cat Swarm Optimization and Attention-Based Graph Convolutional Neural Network for Energy Management Analysis of Grid-Connected Hybrid Wind-Microturbine-Photovoltaic-Electric Vehicle Systems

Abstract

A Hybrid Wind-MicroTurbine (MT)-Photovoltaic (PV)-Electric Vehicle (EV) system integrates multiple renewable energy sources (RES) and storage technologies to optimize power generation, distribution, and consumption. However, the high cost of installing wind Turbine (WT), MT, PV panels, and Energy Storage Systems (ESS), along with the necessary infrastructure, makes it a costly solution, particularly for small-scale or residential applications. To address these challenges, this paper proposes a hybrid approach for the economic assessment of a grid-connected hybrid Wind-MT-PV-EV system. The proposed method combines the Sand Cat Swarm Optimization (SCSO) algorithm with the Attention-Based Sparse Graph Convolutional Neural Network (ASGCNN), forming the SCSO-ASGCNN technique. The goal is to enhance the economic performance, cost-effectiveness, and dynamic control of the hybrid system. The SCSO algorithm is employed to optimize energy management (EM) and improve the operational efficiency of the system, while the ASGCNN is utilized to predict the forecast patterns of energy generation and consumption. The proposed method is implemented on the MATLAB platform and evaluated against several existing approaches, including the Adaptive Genetic Algorithm (AGA), Proximal Policy Optimization (PPO), State-Action-Reward-State-Action (SARSA), Deep Reinforcement Learning (DRL), and Modified Dragonfly Algorithm (MDA). The results show that the SCSO-ASGCNN method achieves the lowest average cost of $532.63, demonstrating its superior performance in minimizing costs compared to other methods.