Abstract: Because the current method does not repair the missing value of power generation data when collecting power generation data,there are some problems,such as poor data acquisition accuracy before missing data repair and much difference from the actual collected data. Therefore,this paper proposes a distributed photovoltaic power station generation data acquisition method based on time series. According to the different effects on the photovoltaic power station,this method constructs a time series model of load abnormal value type,and uses this model to eliminate the abnormal data load points.Because the deleted data has missing values,it is repaired.According to the repair results,the BP neural network is used to collect the power generation data, so as to complete the collection of power generation data. The experimental results show that the method is effective and practical through the comparative test before and after missing data repair,and the data collection test of different methods and actual indicators.