A fuzzy classification method based on rules learning for shaded and unshaded hotspot faults on photovoltaic modules

Abstract

Hotspots (HS) reduce the output power (P) of photovoltaic modules (PVMs) and sometimes lead to fires. However, it is difficult to effectively classify and detect HS faults. In this paper, we focus on the classification and detection of Shaded HS and Unshaded HS. HS faults not only occur due to non-uniform illumination and parameter mismatch but due to internal defects as well, such as; low resistance defects of a cell. The Shaded HS can be repaired sometimes by water washing, such as PVMs in the desert, while the Unshaded HS can only be repaired by replacing them. Therefore, to classify and detect such HS faults, we propose a fuzzy classification method based on data-driven rules learning. Firstly, the characteristic performances of Shaded and Unshaded HS PVMs are compared with the performances of normal PVMs. Secondly, the percentage reduction algorithm is presented for extracting three fault indicators to identify such HS faults. Then, a classification method is proposed using a technique called the Mamdani-type Fuzzy logic system, in which the rules are obtained by a process called the “Data-Driven Rule Generation Method”. Finally, a low-resistance HS model is built using Matlab Simulink. The simulation and experiment results show that the proposed method has more classification accuracy for Shaded and Unshaded HS faults.