Joint Path and Charging Station Optimization for Multiple EVs Based on Improved Ant Colony and Self-Supervised Learning

The rapid development of electric vehicle (EV) technology is driven by energy shortages and environmental concerns. However, EVs face limitations, including the need for frequent and time-consuming recharging to sustain travel. During peak hours, traffic congestion and queuing at charging stations result in EVs spending more time on routing compared to conventional vehicles. This study addresses the EV charging scheduling problem, aiming to minimize the total elapsed time, including charging time, by jointly optimizing charging path routing and charging station selection. To enhance optimization performance, we integrate deep learning and self-supervised learning techniques. Considering the NP-hard nature of the optimization problem, factors such as remaining battery energy, prioritization, traffic conditions, and charging time are incorporated into the proposed method. An EV charging scheduling approach is developed based on the improved ant colony algorithm to identify the optimal charging path. Compared to the traditional ant colony algorithm, the improved ant colony algorithm improves the pheromone update strategy. Simulation results demonstrate that the improved ant colony algorithm significantly reduces total elapsed time compared to the greedy algorithm and the traditional ant colony algorithm. This study offers a practical approach combining artificial intelligence and optimization theory for real-world EV charging management.