Abstract
Introduction. The integration of photovoltaic (PV) systems with electric vehicle (EV) charging infrastructure requires intelligent energy management strategies capable of handling nonlinear dynamics, fluctuating irradiance conditions, and variable load demands. Advanced predictive control techniques are essential to ensure stable, efficient, and adaptive system operation under real-world conditions. Problem. Conventional maximum power point tracking and control approaches often suffer from slow transient response, steady-state oscillations, and reduced tracking performance during rapid environmental changes, limiting their effectiveness in dynamic renewable-integrated charging systems. Goal. To develop a predictive-adaptive intelligent energy management framework that improves dynamic response, tracking accuracy, and overall efficiency for PV-battery integrated EV charging applications. Methodology. A hybrid control architecture combining bidirectional long short-term memory predictive modeling with particle swarm optimization is proposed. The framework integrates real-time data acquisition, signal preprocessing, machine learning-based prediction, and optimization-driven adaptive duty-cycle control to regulate power flow between PV generation, battery storage, grid interaction, and EV load. Results. Simulation results demonstrate significant performance improvements compared with conventional controllers, including tracking accuracy of approximately 99.5–99.7 %, system efficiency of about 98.2 %, reduction in settling time from approximately 0.42 s to 0.18 s, and steady-state ripple reduction from about ±2.5 % to nearly ±0.5 %. Scientific novelty. The proposed work introduces a unified predictive-adaptive framework that integrates deep learning-based temporal prediction with evolutionary optimization, enabling real-time adaptive control in renewable-integrated EV charging systems. Practical value. The developed framework enhances operational stability, energy utilization efficiency, and battery management performance, supporting practical deployment in smart EV charging infrastructure and intelligent renewable energy systems. References 22, tables 4, figures 9.