The type and number of distribution network misoperations are increasing, which brings challenges to the safe and stable operation of the power grid. In this paper, we propose a multi-layer and multi-dimensional intelligent discrimination method based on rules engine and deep learning to prevent misoperation in power grid. Firstly, a rule engine driven misoperation recognition method is proposed, which fully considers the dimensions of command execution, equipment status and personnel operation behavior, and sifts out useless events through a minimum perfect Hash table to improve identification efficiency. Secondly, a hybrid deep learning based multi-layer potential misoperation recognition is proposed, which deeply analysis and processes the historical operation data to identify potential misoperation modes, and also effectively supplements and updates the rule base of the rule engine to enhance the automation level of the system. Simulation results show that compared to the two comparison algorithms, the proposed method can improve the success rate of misoperation recognition by 2.10% and 3.06%, respectively.