The monitoring system of photovoltaic power station yields a large number of abnormal original data in operation,which leads to the deterioration of the overall accuracy and integrity of the data,and then affects the photovoltaic power station to carry out power prediction and condition monitoring.Therefore,this paper summarizes the abnormal detection technology of power data of photovoltaic power stations.Combined with the original power data of grid connected photovoltaic power stations in Ningxia,this paper analyzes the data quality of photovoltaic power stations,and summarizes the types and distribution char-acteristics of abnormal data of actual photovoltaic power stations.According to the three classifications of sampling anomaly a-nalysis,statistical feature analysis and intelligent algorithm analysis,the traditional data anomaly detection methods are ana-lyzed,compared and summarized.On this basis,the combined data anomaly detection method of change point quartile method is discussed,and the view that the combined method has more advantages in data anomaly detection than a single method is pro-posed.The future challenges and development trends in the field of photovoltaic power station power data anomaly detection are prospected.
关键词
光伏电站/功率数据/异常检测/数据重构
Key words
photovoltaic power station/power data/abnormal recognition/data reconstruction
维普
万方数据