AI-OPTIMIZED SOLAR-POWERED FAST CHARGING STATION FOR ELECTRIC VEHICLES WITH IOT-BASED ENERGY MANAGEMENT SYSTEM: A SUSTAINABLE APPROACH FOR SMART CITIES

Abstract

The rapid proliferation of Electric Vehicles (EVs) has created an urgent need for sustainable, efficient, and intelligent charging infrastructure. Conventional grid-dependent charging stations impose significant load on the power grid and rely on non-renewable energy sources. This research proposes a novel AI-Optimized Solar-Powered Fast Charging Station integrated with an IoT-Based Energy Management System (EMS) for smart cities. The proposed system integrates a 50 kW monocrystalline solar PV array with a bidirectional DC-DC converter, Maximum Power Point Tracking (MPPT) using Perturb & Observe and Artificial Neural Network (ANN) algorithm, and a 100 kWh Lithium-ion Battery Energy Storage System (BESS). An IoT platform based on ESP32 and cloud integration (ThingSpeak/Blynk) enables real-time monitoring of solar irradiance, battery State of Charge (SoC), charging demand, and grid load. The core innovation is the development of a Deep Reinforcement Learning (DRL) model that optimizes energy flow between PV, BESS, EV, and Grid (V2G/G2V) based on weather forecasting, traffic pattern, peak-hour pricing, and load prediction. MATLAB/Simulink and HOMER Pro simulations were used to validate the system under Indian climatic conditions (Solapur region data). Results demonstrate a 34.7% improvement in charging efficiency, 28% reduction in grid dependency, and 41% reduction in charging cost compared to conventional grid-based stations. The system achieves a fast-charging capability of 80% SoC in 45 minutes with an overall system efficiency of 96.2%. The proposed solution offers a scalable, cost-effective, and environmentally sustainable charging infrastructure contributing to net-zero mobility goals.