Intelligent DC Bus Voltage Quality Enhancement in Solar-Powered EV Charging Infrastructure Using Temporal Fusion Transformers with Self-Attention Mechanism

Electric vehicle (EV) charging stations contribute to sustainable energy growth; however, they face issues with poor power quality (PQ) and harmonics due to DC bus voltage instability caused by variable loads, fluctuating solar output, and nonlinear charger effects. To overcome these limitations, this research presents an intelligent framework for improving PQ in solar-powered EV charging systems. To guarantee effective solar energy extraction, the system architecture comprises of a DC bus integrated with a bidirectional DC–DC converter that uses Maximum Power Point Tracking (MPPT). Harmonic distortions and voltage ripple are introduced by modelling the EV charger as a nonlinear load. A shunt Active Power Filter (SAPF) is attached to the DC bus to minimize these disruptions. A novel control strategy is introduced using a hybrid forecasting model, Temporal Fusion Transformer with Self-Attention Mechanism (TFT-SAM), which predicts Total Harmonic Distortion (THD) based on input features from the DC bus. The error between predicted THD and predefined reference (THD_ref) is used to generate optimized Pulse Width Modulation (PWM) signals for the SAPF, enabling dynamic harmonic compensation and voltage ripple reduction. MATLAB simulations show that the proposed TFT-SAM achieves 99.40% efficiency, minimizes THD to 0.60%, and provides a quick response time of 25 ms, ensuring dependable and superior EV charging.