Photovoltaic fault diagnosis method based on BP neural network and rule-based control chart

Abstract

PV (Photovoltaic) array is the fundamental power production unit that is installed in outdoors and continuously exposed to harsh environmental conditions. This inevitably leads to abnormal and various kind of faults among PV arrays at times. Therefore, the accurate detection of potential compound faults is essential to improve the operating efficiency, safety and reliability of PV systems. In order to monitor the operating state of PV arrays in a real-time manner, a fault diagnosis method based on BP (Back Propagation) neural network and rule-based control chart was proposed. A DC output prediction model of PV arrays was constructed using BP neural network. The actual output deviation of PV arrays was analyzed based on the predicted output of the model, and the EWMA (Exponentially Weighted Moving-Average) control chart was used to detect whether the actual output deviation of PV arrays exceeds the EWMA control limit to further determine whether a fault occurs. The proposed method builds a prediction model by taking advantage of the environmental data collected by micro-weather station and the measurement data of PV inverters for the remote real-time online monitoring and fault diagnosis of PV arrays without any additional monitoring equipment.