Research On False Data Injection Attack Mehods And Mechanisms Of Defending In Smart Grid

All aspects of social life are inseparable from the stable operation of the power network.The stability of the power network is related to the development of the country and society.In recent years,with the development of information technology,the combination of power grid and information technology has become closer.The concept of smart grid generated and continuously developed,and it is a vital development direction of power systems.With the developing of information technology,the security problems of smart grid have gradually increased,and the attack method have also diversified.Among the many attacks on smart grids,false data injection(FDI)attack poses a huge threat to the security of smart grid.The FDI attack makes the data transmitted to the control center which are different from the measurement data by injecting data into devices such as sensors in the smart grid,thereby affecting the evaluation result of the control center to estimate system state and making the control decision error.It has been proposed that the false data detector based on Support Vector Machine(SVM)has good detection effect.The false data detection module is located in the energy management system of the control center,which is called the central detection model.The process of false data detection can also be performed in the remote sub-controller,which is called the edge detection model.Based on these two detection models of the smart grid,this paper proposes an FDI attack method based on hyperplane migration and its corresponding detection method.The core of the FDI attack method based on hyperplane migration is to construct an attack vector that is spatially located in the positive sample space region adjacent to the hyperplane of the SVM.Then,these attack vectors are injected into the sensors,and the sensors transmits the data to the control center through the SCADA system.After the data is detected,the control center updates it to the training set and it will affect the next training result,thereby achieving the attack effect.The FDI attack method based on hyperplane migration is applied to the smart grid based on the central detection model and the smart grid based on the edge detection model.(1)In the smart grid based on the central detection model,this paper proposes a detection method based on data distribution and a detection method based on association analysis.The data distribution based detection method refers to detecting whether there is an attack by comparing the distribution of the training centralized data with the newly received data of the control center;the detection method based on association analysis refers to the change of the data through the training and the external environment of the power grid.(2)In the smart grid based on edge detection model,this paper proposes a detection method based on node analysis and a detection method based on data distribution.The detection method based on the node analysis refers to detecting the attack by comparing the differences of the data in different sub-controllers;the detection method based on the data distribution is the same as the central detection model.In order to verify the proposed attack and detection method,the simulation experiment environment used in this paper is MATLAB,and the simulation experiment will be carried out in the IEEE-14 bus,IEEE-39 bus and IEEE-118 bus standard test systems in the Matpower package.The experimental results show that the attack method we proposed in this paper has achieved the corresponding attack effects under both the central detection and edge detection models.At the same time,the detection method we proposed in this paper can effectively detect the FDI attack.In summary,this paper mainly studies the FDI attack methods and defense methods in the smart grid.An attack method and the corresponding detection methods are proposed for the false data detector based on SVM.The possible attack methods and new defense ideas are provided for the research on the false data detector based on machine learning,which has a high theoretical and practical value.