Deep-learning based power switch fault diagnosis in DC/DC converters for photovoltaic applications

This paper presents a robust approach to handle with the diagnosis of open-circuit and short-circuit faults in a DC/DC converter used for PV applications. The proposed approach uses only the measurements sent by the PV current and voltage sensors. Then after, a deep learning (DL) based classifier is built to detect the power-semiconductor fault and to discriminate open-circuit fault from short-circuit fault. In this work, a comparison between KNN, SVM, LSTM and BiLSTM models is discussed. Several simulations under MATLAB/Simulink software are presented to illustrate the effectiveness of the proposed approach.