Application of Sparse Identification of Nonlinear Dynamics (SINDy) Towards Modeling and Prediction of Potential Induced Degradation (PID) Susceptibility of Photovoltaic Modules

The process of Potential Induced Degradation (PID) occurs when a potential difference exists between the photovoltaic cells and the outer part of the modules, resulting in leakage currents and rapid temperature rise, eventually resulting in system failure and power losses. Using Sparse Identification of Nonlinear Dynamical Systems (SINDy), the paper developed a PID Susceptibility Prediction Model for photovoltaic (PV) modules. Our goal is to elucidate the nonlinear dynamics and interactions among key parameters contributing to PID susceptibility. PID susceptibility was determined to be significantly affected by temperature, relative humidity, voltage bias, and leakage current based on a thorough literature review. The model captures the nonlinear dynamics and interactions among the factors contributing to PID susceptibility with an MAE ≈ 0.35 and RMSE = 0.40. Temperature, relative humidity, voltage bias, and leakage current all significantly affected PID susceptibility. Temperature and relative humidity have the greatest influence, but their interactions are equally important. Voltage bias has a relatively small effect on PID degradation susceptibility, but it cannot be ignored. Experiment results demonstrate substantial proof to support the effectiveness of Sparse Identification of Nonlinear Dynamics (SINDy) to capture system dynamics and provide valuable tools for accurate predictions.