Aiming at the abnormal state in the operation process of the control system, a detection method of abnormal data in the operation of smart grid control system based on rough entropy weighted density is proposed. The rough set is used to analyze and reason the operation data of power grid control system. The complex uncertain relationship is transformed into Boolean data and sorted by using the cut relationship method of membership degree. The abnormal data in the operation of smart grid control system are detected based on the weighted density of objects, and the abnormal data of various functions of the control system are accurately identified. The proposed method is verified by the real power grid control system data, and the results show that the proposed method has higher accuracy and lower omission rate compared with the traditional abnormal state identification method.