PV System Faults Detection and Classification Using Multi-class Support Vector Machine Algorithm

In this article, an in-depth analysis was conducted to solve the significant challenge of early and accurate fault detection in PV systems. In the first step, the modeling of a PV system in healthy mode and degraded mode is carried out. After this, five types of defects were considered and implemented in order to form a database for diagnosis purposes. In the second step, a diagnostic approach based on a multi-class Support Vector Machine algorithm is proposed. This approach, using an RBF (Radial Basis Function) kernel and a multiple binary classifier, has shown its high ability and efficiency to detect and classify defects in a PV system.