Health status identification method of PV plant sensors based on abnormal data analysis

Abstract To achieve accurate control of the PV plant, various sensors are deployed. Identifying the health status can improve the safety of PV plants and the accuracy of PV power generation prediction. To enhance the accuracy of sensor fault type identification and make the process interpretable, a method for identifying the health status of PV plant sensors based on abnormal data analysis is proposed. The analysis including general fault types and abnormal data of PV plant sensors is analyzed. A model for detecting the type of abnormal data behavior is established, and fault diagnosis is performed based on the abnormal data behavior to determine the specific faults of the sensors. Experimental verification is carried out using data from three sensors: temperature, humidity, and light intensity in a PV plant. The experimental results show that the method can accurately and comprehensively identify the health status of PV plant sensors, with an accuracy of 99.53%, and an F1-score of 0.9825.