A novel reconfiguration strategy for PV system under mismatch irradiance with experimental validation and machine learning-driven fault detection and diagnosis

The photovoltaic systems are increasingly adopted as sustainable energy sources, yet their performance is severely affected by partial shading. Reconfiguring the connections of the PV array is an effective way to reduce the negative effects of partial shading. Therefore, motivated by various reconfiguration techniques available in the literature, this study contributes a novel PV array reconfiguration strategy based on Snake Optimizer algorithm. A 9×9 PV array demonstrates that the proposed Snake Optimizer consistently outperforms the conventional total cross-tied, Improved Northern Goshawk Optimization, and Triple X Sudoku methods in terms of overall performance and effectiveness. Furthermore, to ensure reliable operation under hardware degradation, a machine learning framework is also employed for fault detection, diagnosis, and classification. Additionally, a 3×3 experimental testbed validates the proposed Snake Optimizer-based reconfiguration in real-world conditions, confirming its effectiveness in mitigating shading effects. In order to show the efficacy of the proposed Snake Optimizer, it is employed for the symmetrical, asymmetrical PV array and scalable to larger photovoltaic system. It is observed that it offers a practical and robust solution for improving PV energy harvesting under non-uniform irradiance.