Condition monitoring of a large-scale PV power plant in Australia

This paper considers the problems of condition monitoring and fault detection in an existing solar photovoltaic (PV) plant in Australia. A PV prediction model is proposed to accurately model the PV plant output. This model is then used with three condition monitoring and fault detection methods. The considered methods involve comparison of measured and modeled voltage and current ratios with appropriate thresholds and adjacent string values. The procedure to calculate thresholds is described. The proposed PV model and the condition monitoring approaches are collectively used with real PV data. As a result, the string disconnection and bypassed module faults are detected. It is also found that the string level monitoring is ideally suited for reliable condition monitoring and fault detection especially for large PV plants.