Research On False Data Detection Of Power System Based On Static Network Representation

With the expansion of smart grid and the continuous research on smart grid,it has become a major hidden danger for power system security for hackers to attack information for illegal interests.The research shows that the false data injection attack can bypass the bad data detection of traditional state estimation,and there may already be false data in the historical data of power system.At the same time,in order to provide normal prior data for subsequent real-time data detection,it is necessary to detect the false cross-section data in the historical data.Therefore,based on the background that the false data injection attack avoids the traditional state estimation detection,this paper deeply studies the static network detection method of power system.The details are as follows:Firstly,aiming at the problem that the false data injection attack avoids the traditional state estimation detection,it is explained that the traditional power system state estimation is divided into AC and DC,and some attack processes of the false data injection attack are deeply studied,and a method of simulating False data injection in DC state is proposed.Based on the measurement data of IEEE-118 bus and IEEE-2383 bus systems used in this paper,this method establishes the false data injection attack model,which lays the foundation for establishing the false data injection attack detection model.Secondly,a static false data detection method based on the similarity characteristics of network nodes at a certain moment is proposed,and in order to better represent the power system in complex networks,this method combines the structure and attributes of nodes in the power grid,and uses the egonet model of the power grid to represent nodes,making the false data injection attack detection model more intuitively and clearly.A scheme of simulating cyclic iteration is designed to determine the classification number of nodes in order to improve the accuracy of clustering results.The inherent properties of nodes and cross-sectional data(including normal data and abnormal data)are treated separately to solve the problem of outputting the false data injection attack test results,and classification clustering is used to distinguish false data.The simulation results show that this method is effective and can detect more than 80% of the possible false data injection problems in the power system.Finally,the specific significance of power flow direction to the directed graph of power grid is excavated,a static network representation of static false data considering the flow direction of active power flow is proposed,and the out-degree and in-degree matrices are constructed to make the model closer to the actual power system.In the process of using affine propagation clustering algorithm to solve the problem,the influence of numerical setting of key parameters on the calculation results is taken into account,and the parameters are set within the allowable range in order to select more suitable parameters.This method is tested on IEEE-118 bus and IEEE-2383 bus systems,and the detection results obtained by affine propagation algorithm show the applicability of this method.