ANN based energy management of solar integrated bidirectional EV charging stations for light electric vehicles in Arba Minch Ethiopia
摘要
The integration of solar photovoltaic (PV) generation with bidirectional electric vehicle (EV) charging infrastructure provides a promising solution to reduce grid dependency and support sustainable electric mobility. However, the intermittent nature of PV generation and the state-of-charge (SOC) constraints of battery storage introduce challenges in real-time energy coordination among PV, standby battery, EV load, and the utility grid. This paper proposes an ANN-based energy management system (EMS) for a solar-integrated bidirectional charging station for light electric vehicles (LEVs), with Arba Minch, Ethiopia considered as a representative emerging-grid case study. The proposed approach introduces an SOC-aware ANN controller that enables adaptive coordination of PV, battery, and grid power under varying irradiance and load conditions. The ANN is trained using PV power, EV demand, and battery SOC as inputs to predict the required battery power reference for stable system operation. Simulation results demonstrate effective ANN learning and control performance, with the regression coefficient approaching unity and stable convergence of the training process. Under step irradiance variation from 1000 to 200 W/m2, the system maintains a regulated DC-link while sustaining a constant 30 kW EV charging demand. The proposed EMS dynamically allocates power among PV, battery, and grid sources, with SOC-dependent operation ensuring battery protection and reliable load support. The results confirm that higher SOC conditions reduce grid dependency, while lower SOC conditions increase grid support during PV deficit. Overall, the proposed ANN-based EMS provides a reliable and adaptive framework for coordinated energy management in renewable-powered bidirectional LEV charging systems and demonstrates strong potential for deployment in emerging-grid environments.