Photovoltaic (PV) systems are frequently subjected to several faults leading to costly production losses. The proposed work focuses on the improvement of the energy efficiency of a DC micro-grid by minimizing the losses related to the occurrence of PV faults. First, we presented a state of the art on the most recurrent faults and their diagnosis methods. The literature review has led us to adopt a data-driven diagnosis approach. Then, the Principal Component Analysis (PCA) method was proposed for PV shading fault detection and classification for a PV module of 250 Wp. The PCA was first performed using the entire I(V) curve obtained under real climatic conditions. A minimum classification success rate of 87.38% is obtained in the training s...