Artificial Intelligence and Optimization Techniques for Dual-Stage Interleaved Electric Vehicle Onboard Chargers: A Review

Abstract

The rapid electrification of transportation has increased the demand for efficient, compact, and intelligent onboard charging systems for electric vehicles (EVs). Dual-stage interleaved onboard chargers have emerged as a promising solution due to their high power density, improved efficiency, and reduced harmonic distortion. However, conventional control strategies struggle to manage nonlinearities, dynamic load variations, and system uncertainties. This review explores advanced control and optimization approaches for dual-stage interleaved onboard chargers, focusing on the integration of PIDD2-PD controllers and deep learning techniques. The PIDD2-PD controller enhances transient response, robustness, and stability, while deep learning models enable predictive control and real-time optimization of nonlinear system behavior. Additionally, hybrid metaheuristic optimization methods, particularly the Hybrid Adaptive Genghis Khan Shark Gold Rush (HAGK-SGR) algorithm, are analyzed for efficient parameter tuning and multi-objective optimization. These approaches improve convergence speed, global optimality, and system performance. Applications across residential charging, grid-connected systems, and vehicle-to-grid scenarios are examined using simulation and real-world datasets. Performance metrics such as efficiency, power factor, and total harmonic distortion are evaluated. Findings indicate that hybrid deep learning and optimization frameworks significantly enhance efficiency, adaptability, and reliability, supporting next-generation intelligent EV charging systems.