Establishing a detection model data attacks in power distribution system

Abstract

The safe operation of smart distribution network is highly dependent on the powerful technical guarantee provided by the function of information link, which makes the network vulnerable to the threat of malicious data injection and other network attacks during the operation. In order to ensure that this kind of malicious data injection attack can be detected sensitively in the operation of power grid, this paper proposes a kind of power system state estimation malicious data attack defense model based on historical data. Firstly, the Long Short-Term Memory(LSTM) network is trained with the historical state quantity to realize the state prediction model. The prediction results are used as a reference, and the deviation between the prediction and the real-time estimate is calculated to break the concealment of malicious data. Simulation results of IEEE33-bus power system verify the accuracy of prediction and the effectiveness of the proposed method for online detection of hidden malicious data.