Fault Detection and Classification in Solar Photovoltaic Array

Solar photovoltaic (PV) technology will undoubtedly meet future power demand directly or indirectly. Solar PV modules are combined to form larger solar arrays. In recent years, the rating of solar PV systems has increased, as the number of conversion units introduced to photovoltaic installations. It also increases the risk of failure at any power plant stage. Because solar panels are energized by sunshine throughout the day, it is difficult to shut them down during failures on the DC side of PV panels. Increased voltage and current rating of Solar PV modules causes increase in the possibility of huge fault currents or DC arcs, which may result in decreased system performance, accelerated system depreciation, and fire hazards. This paper presents a neural network-based fault detection and classification system. This approach enables the early diagnosis, stratification and positioning of the PV array faults and also helps to determine the fault's type. The results showed that the wide neural network provided the highest accuracy and lowest computation time.