The invention provides a data processing method for a photovoltaic power station data acquisition system. The method is characterized by comprising the steps of firstly, subdividing data into normal data and abnormal data, and dividing the abnormal data into ones caused by a human reason and a system reason by analyzing and verifying the abnormal data; correcting the abnormal data caused by the human reason by judging a threshold, and directly deleting the data which cannot be corrected; correcting the abnormal data caused by the system reason by a variant joint matching method, and directly deleting the data which cannot be corrected; and then combining the normal data with the corrected data, processing missing values, dividing the missing values into random and non-random missing values according to a relation between the missing values and incomplete variants, and filing the missing values with a calorie filling method or a polynomial filling or mean filling method respectively. The method not only can improve the quality of data, but also can improve the reuse value of data.