Robust artificial neural network-based control for photovoltaic integration under grid fault conditions

This study presents a robust artificial neural network (ANN) controller designed for a two-stage voltage source converter in grid-connected photovoltaic systems. The core innovation of this work is the comprehensive application of ANN across the entire system, including DC bus stability and grid power control. Developed and integrated within the MATLAB/Simulink environment, the controller aims to improve dynamic response and robustness against irradiance variation, external noise disturbances and grid faults. Comparative analysis against integral terminal sliding mode control (ITSMC) and traditional sliding mode control (SMC) reveals the ANN controller superior performance. Indeed, the proposed ANN controller demonstrates a Total Harmonic Distortion (THD) of 0.58% , outperforming the ITSMC and SMC. This ensures compliance with the IEEE Std 519–2014 requirements. Additionally, the ANN controller exhibits a significantly faster dynamic response, reducing the settling time by a factor of more than 6. The ANN controller also excels in minimizing tracking errors and overshoot, while maintaining robust performance under noisy and faulted grid conditions, as confirmed by integral performance indices.