Automatic Fault Detection of Photovoltaic Modules Using Recurrent Neural Network

Abstract

Everywhere in the globe, the total capacity of photovoltaic (PV) panels is expanding at an exponential rate. Arc faults, open-circuit (OC) faults, bypass diode failures, mismatch faults, and short circuit faults are only a few of the most common types of problems that may occur in PV arrays. Not recognizing and correcting these issues quickly might affect power plant production. Fault detection in PV modules helps stabilize PV plant output. Machine learning techniques can automatically identify PV module issues. This paper portrayed fault detection using Multi-Layer Perceptron (MLP) and Recurrent Neural Network (RNN). Two methods identify PV defects based on normalizing factors. MLP has nonlinear problems and is slow to compute. The suggested RNN proved to be a superior detection approach for 10 weeks of testing on 2.4 KW monocrystalline solar panels. MLP has 75.62% fault detection accuracy whereas RNN has 98.95% in 4s-2p PV panels. Therefore, the findings of the simulation indicate that the proposed RNN technique achieves the necessary level of speed and accuracy.