Distributed State Estimation Of Active Distribution Network Considering False Data Injection Attack

With the continuous development of solar power generation and wind power generation technology,more and more distributed power sources are connected to the distribution network.The traditional single power distribution network is gradually changing into multi power active distribution network.Compared with the traditional distribution network,the control and operation of active distribution network are more complex,the amount of transmission information is large,it is difficult to avoid bad data,topology errors,which is easy to cause inaccurate state estimation.The randomness and intermittence of new energy also interfere with the identification of bad data.In addition,compared with the large power grid,the network composition of the active distribution network information and communication system is diverse and mixed,the network access authority is relatively open,and the security defense measures are far less complete than the large power grid information and communication system.This makes it easy for hackers to obtain the relevant system measurement configuration and real-time measurement information,and construct the attack vector of malicious injection of bad data into the power grid,resulting in a serious threat to the system data security.Therefore,it is of great significance to study the state estimation suitable for the development of active distribution network and the detection and defense methods against false data injection attacks.Considering the false data injection attacks under the nonlinear model,a multi region distributed state estimation method based on the average consistency algorithm and a false data injection attacks detection and bad data correction method based on neural network are proposed for the active distribution network.The specific work is as follows:1.The principle of power system state estimation and false data injection attack are introduced,and the construction method of false data injection attack based on nonlinear state estimation is analyzed.2.Considering large measurement error and vulnerable to false data attack of active distribution network,a multi-area state estimation method for active distribution network is proposed.Each subsystem exchanges information through the average consistency algorithm,configures PMU at the vulnerable node for protection,and provides voltage amplitude and phase angle measurement for the subsystem.3.A false data detection and correction method based on neural network is proposed.The residual detection and fully connected neural network are used to judge whether the system is attacked by false data injection,and the multivariable LSTM model is used to predict and correct the bad data.Simulation results show that the proposed method can effectively identify false data injection attacks and correct bad data.