Electric Vehicle Parking Lot Scheduling Using Parallel Genetic Algorithm on a Graphics Processing Unit

Electric Vehicles (EV) are increasingly popular due to environmental benefits but can strain the power grid. Optimizing EV charging in large parking lots is a challenge. This study uses a metaheuristic approach on a graphics processing unit (GPU) to schedule EV charging in a smart parking lot. The proposed method uses a genetic algorithm (GA) to create an efficient charging schedule that minimizes cost and respects charger and feeder capacities. The algorithm leverages parallelism in GPU. Tested on lots with different vehicle capacities, the proposed GA-based method is compared to a particle swarm optimization (PSO) method and achieve better results. The GPU implementation speeds up the calculation by 134.5x compared to a sequential execution on CPU. It can optimize a 20-EV lot in 0.46 seconds and a 500-EV lot in 5.80 seconds, allowing for real-time computation.