Multiple-Regression method for Online Fault Detection and Diagnosis of PV Systems Using Kalman Filter Algorithm

Faults on photovoltaic (PV) systems can drastically degrade Microgrids reliability, and stability, if not promptly detected. Thus, a novel Fault Detection and Diagnosis (FDD) methodology is proposed for online monitoring of PV system DC side. This method is based on multiple non-linear regression to emulate the PV behavior at different weather conditions precisely. The regression method is formulated on the relationship of PV characteristics at the maximum operating point without irradiation sensors. Then, to restrain uncertainties and measurement noises, a Kalman Filter algorithm is used. In addition, an adaptive threshold based on non-linear polynomial regression is developed to detect early fault signature in a PV system. To evaluate the performance of the proposed FDD approach, short circuit fault is investigated via MATLAB/Simulink® at various weather conditions. The result reveals the effectiveness of the proposed FDD method to detect soft faults even at low irradiation.