Adaptive Grid‐Vehicle Energy Management Using Goat Optimization and Convolutional Kolmogorov‐Arnold Networks With Hybrid Renewable Energy Integration

ABSTRACT

Hybrid renewable energy sources (HRESs)‐based grid‐vehicle integration enables coordinated vehicle‐to‐grid (V2G)/grid‐to‐vehicle (G2V) interactions between renewable generation units, the grid, and vehicle batteries, supporting flexible and balanced energy exchange under continuously changing operating conditions. To overcome these limitations, this manuscript suggests an efficient approach for adaptive V2G and G2V control in grid‐vehicle integration with HRESs. The suggested method combines the goat optimization algorithm (GOA) with the convolutional Kolmogorov‐Arnold network (CKAN), referred to as the GOA‐CKAN technique. The key objective is to reduce operating cost, enhance energy exchange efficiency, minimize power loss, lower emissions, and maintain stable bidirectional power flow throughout V2G/G2V processes. GOA optimizes the bidirectional power scheduling between the grid and vehicles by adjusting charging and discharging decisions for stable energy exchange. CKAN predicts the required power flow patterns during V2G/G2V operation based on varying renewable inputs and system conditions. The method is implemented on MATLAB and compared with several existing approaches such as artificial neural network (ANN), pelican optimization algorithm‐triple‐Memristor Hopfield neural network (POA‐TMHNN), ANN‐particle swarm optimization (ANN‐PSO), adaptive interaction ANN (AI‐ANN), and Quantum Neural Network (QNN). The GOA‐CKAN method achieves a total operational cost of $1563, minimizes power loss to 15.6 kW, improves efficiency to 99.2%, and reduces emissions to 60.6 kg, thereby demonstrating superior capability to optimize bidirectional energy exchange, regulate EV charging and discharging, and ensure sustainable and reliable V2G/G2V operation within grid‐vehicle systems integrated with HRESs.