Data-driven approach for isolated PV shading fault diagnosis based on experimental I-V curves analysis

This paper deals with a data-driven fault diagnosis method for photovoltaic (PV) system. The proposed method is based on the Principal Component Analysis (PCA) to detect and identify different shading types. The PCA uses the current-voltage (I-V) curves that are experimentally determined for a monocrystalline PV module of 250Wc. The experimental tests were carried out for several shading patterns covering the PV cells. For the diagnosis process, three features (current, voltage and power of PV module) are extracted for each test to build the database which is then analyzed through the PCA algorithm. Simulation results using the experimental data, prove the efficiency of the proposed method in terms of discrimination. The healthy data are clearly separated from the faulty ones despite sudden irradiation variations.