Research On Security Detection Of Smart Grid Under False Data Injection Attacks

With the development of automatic control technology and network communication technology,the integration of power system information network and physical equipment is gradually deepening,and it is shifting to the direction of intelligence.The development of smart grids is essential to efficient power transmission and distribution,dispatch management,economic operation,etc.However,information security problems of networked devices occur from time to time.False Data Injection Attacks(FDIAs)is a common network attack method.Attackers inject false data information through smart grid sensors,controllers,and communication networks,tamper with the original data of the grid,and then affect control decisions.Causes the malfunction of power grid equipment,and even paralyzes the power network,threatening the security of the power grid and causing huge economic losses at the same time.The current research on FDIAs in smart grids is still in the preliminary stage.Because of its hidden features,traditional attack detection methods are difficult to detect.Therefore,it is of positive significance to study FDIAs detection in smart grid.Aiming at the security detection problem of smart grid under false data injection attacks,this paper mainly carries out the following work:1.Establish a power system model and construct FDIAs vectors on the basis of state estimation.First,the smart grid system is outlined.Considering the different forms of attacks by attackers,FDIAs in smart grids are divided into three categories.Secondly,the importance of state estimation in the smart grid is explained,and a system model is established for the non-linear characteristics of the grid system.A state estimation method based on Unscented Kalman Filter(UKF)is proposed to improve the accuracy of power system state estimation,and simulation experiments are carried out in the power IEEE-14 node system.Finally,on the basis of state estimation,the principle of FDIAs is analyzed,and the attack vector is constructed,and how the attacker can realize the attack on the power grid based on the information of the power grid topology and other information.2.On the basis of the state estimation of power grid nodes,a detection method of FDIAs based on state estimation is proposed.Aiming at the nonlinear and high-dimensional characteristics of the power grid system,the Square Root Unscented Kalman Filter(SRUKF)algorithm is introduced to improve the stability of the system and the accuracy of state estimation.Combined with the weighted least squares(WLS)state estimation in the power grid system,Euclidean distance is used as the basis for state consistency detection,and the detection threshold is set.Compare the relationship between the Euclidean distance of the state and the detection threshold to realize the online detection of FDIAs.The simulation experiment is carried out on the power IEEE-14 node system,and the results show that the proposed method can effectively detect the attack.3.A FDIAs detection method based on Gaussian Mixture Model(GMM)clustering is proposed.Considering that the state of the grid system nodes obeys Gaussian distribution,and the attack vector constructed by the attacker has similar distribution characteristics and obeys different Gaussian distributions,it is considered that the system measurement data constitutes a GMM.The sensor data collected by the power grid system is divided into two types: normal and abnormal.Construct positive and negative sample data,and solve the parameters of each model component of the mixed model through the generated training set data.The obtained model is used to classify the test set data in two ways,so as to distinguish whether the measurement data has been maliciously tampered with by an attacker,and achieve the purpose of attack detection.It was verified in the power IEEE-18 and IEEE-30 node systems,and compared with the FDIAs detection method based on support vector machines.The simulation results show the effectiveness of the method.