PV array Fault Classification based on Machine Learning

Abstract

The rapid development of photovoltaic power within a few years has added an abundant source of clean energy to the power system of Viet Nam, especially the power shortage become more serious in recent years. Clean energy such as PV power helps to reduce greenhouse gas emissions and becomes a global trend. During operation, the PV array can get into fault conditions which affect the system performance if not detected correctly and timely. Traditional PV fault detection methods such as statistical signal processing, power loss analysis, voltage and current measurement have been used to detect and locate fault position. However, these methods accuracy may be affected by installation conditions or PV array materials. In this research, we propose to use machine learning models to detect and classify faults in PV array based on I-V data. The results show that the machine learning models can detect the fault in the PV array with an accuracy of up to 99.74%.