Research On False Data Injection Attack And Detection Technology Of Smart Grid

With the continuous development of communication technology,network technology and automation control related technology,many physical devices in the power system are gradually integrated with the communication network,and the power grid has already transitioned from the traditional physical grid to today's smart grid.The continuous development of smart grid is of great significance for efficient power transmission and distribution,intelligent power dispatching and economic operation of power grid.However,due to the high coupling between the basic physical facilities and information communication equipment in the power system,the power system is very vulnerable to network attacks.Attackers use external network equipment to connect the power grid,tamper with important data in the power grid,or interfere with and destroy power dispatching decisions,which will do great harm to the power grid.False data injection attack is a kind of typical network attack in smart grid.The attacker injects false power information into the power grid,such as sensors,remote terminal control units or communication equipment in the power grid,tampers with the original data of important nodes in the power grid,affects the power dispatching decision,causes the misoperation of power grid equipment,and leads to security accidents or even paralysis of the power grid.It can be seen that the false data injection attack poses a great threat to the safe and efficient operation of smart grid.Therefore,the research on false data injection attack and attack detection and defense technology is of great significance to the safe,stable and efficient operation of smart grid.Focusing on the research of false data injection attack and attack detection,this paper mainly carries out the following work:(1)This paper studies the general attack model of exchange false data injection.Considering the different degrees of power data and topology information mastered by the attacker,two types of FDIA vectors are constructed based on the model of false data injection attack(FDIA).One is the global data injection attack when the attacker grasps the global information and topology scenario of the power grid,and the other is the local data injection attack when the attacker grasps the local power information and some important topology information,The effectiveness of the proposed two AC FDIA models is verified by simulation experiments on IEEE14 bus system.(2)A region division method based on node importance is proposed.Taking the node importance as a new control distance,combined with the nonlinear least squares state estimation algorithm,and using the mapping partition and clustering algorithm to partition the power grid.The false data injection attack is carried out on the nodes in the region based on the region division.Through the relevant simulation experiments in the IEEE node system,it is proved that the region division algorithm enables the attacker to successfully implement FDIA at the lowest possible cost.(3)A false data injection attack detection method based on machine learning is proposed.Random forest classifier and XGBoost algorithm are used to detect and classify false data injection attacks.A large number of node state variable sample data are used to train the above two classifiers,and the parameters of the classifiers are optimized,so as to effectively improve the recognition accuracy of the two algorithms for false data injection attacks in different scenarios.