| The power system composed of power generation,transmission,transformation,distribution and consumption provides power for all walks of life.In recent years,China's power demand continues to rise.With the in-depth development of smart grid,the traditional power system has evolved into a new form,and the modern power system has developed into a power information physical system CPS,which integrates power network and information network.At the same time,there are some technical loopholes in the process of the development of power information physical system,which make the lawless seize the opportunity to cause damage to the power system.The false data injection attack can make use of the bad data detection vulnerability in the energy management system,maliciously tamper with the state estimation results,and seriously endanger the safe and reliable operation of the power system.At the same time,distribution network has more potential network attack threats because of its complex network topology and low measurement redundancy.In order to ensure the safe and stable operation of power grid,under the condition of deep coupling between information side and physical side of power CPS,it is of great significance to study a set of efficient and feasible false data detection method for power system.In this paper,the attack and detection methods of microgrid under false information injection attack are studied.Firstly,the mechanism and modeling of power information physical system are studied,the state estimation and bad data detection in the system are analyzed,the principle of false data injection attack is deeply studied,the detection principle of false data injection attack is analyzed,the modeling and simulation analysis of false data injection attack is carried out,and the false data injection attack is quantified The loss of power system caused by lightning stroke.Secondly,an improved load redistribution method is proposed.This method transforms the traditional load redistribution problem for non-zero nodes into the flexible load distribution problem for microgrid.Under the condition of bad information detection of power system,based on limited network attack resources,an improved load redistribution attack model is constructed under the framework of microgrid IDR.By changing the translatable load parameters of power system,the calculation of power price cost is misled,which makes the system make wrong decisions and increase the generation cost.Through the attack model experiment based on Microgrid,based on the From the perspective of attackers,the maximum loss attack scheme and the most cost-effective attack scheme are given.At the same time,the simulation results show that the improved load redistribution attack method proposed in this paper is more harmful than the microgrid,which provides a reference for the power grid to defend against such attacks.Finally,for this kind of false data injection attack,an attack detection method based on reliff and improved isolated forest algorithm is proposed.Firstly,the distributed event triggering mechanism is used to detect the attack,then the relief algorithm is used to select the features of the measured data,and finally the improved isolated forest algorithm is used to detect the measured data.The improved isolated forest algorithm introduces the secondary training process to improve the height weight of the isolated tree with high feature vector weight and reduce the height weight of the isolated tree with low feature vector weight,so as to improve the classification ability of the whole classifier.The simulation results show that the attack detection method based on the improved isolated forest algorithm can effectively identify the false data injection attack,and the accuracy is higher than that based on the traditional isolated forest algorithm and the attack detection method based on the long-term and short-term memory network,which verifies the accuracy and superiority of the model designed in this paper. |
PDF Full Text Request