Investigation Of Power Systems Security Defense Considering False Data Injection

Defending against malicious attacks and maintaining the safe and stable operation of power systems has always been an important research topic in the field of power systems.False data injection attack(FDI attack)is a new form of attack that destroys the reliability of power system operating data in recent years.False data injection attacks can take advantage of the loopholes in the existing power system data security defense measures to bring serious threats to the safe and stable economic operation of the power system,mainly reflects in:False data injection attacks have flexibility and concealment when destroying the stability of power systems.Traditional power system security detection methods are difficult to achieve effective defense? False data injection attacks can attack the power system infrastructure to achieve illegal profits.Traditional power system economic security analysis does not consider abnormal economic activities of false data injection attacks? The diversity of false data injection attack forms brings more complex game scenarios to the power system security attack and defense game.Traditional power system security defense strategy decision-making method is no longer applicable.In recent years,research on false data injection attacks has made some progress.However,due to the late start of related research and the complex security defense situation of new power systems,the existing research work is still insufficient.To this end,under the funding of the National Natural Science Foundation of China(51977033)and National Key Research and Development Program of China under Grant 2018YFB1500800,this paper aims at some gaps in the existing research,combined with the attack characteristics of false data injection and the security situation of the new power system,and carries out research work from the following aspects:1.False data injection attack behavior analysis method and online attack detection method based on attacked system mechanism analysis.Aiming at the scenario where the power system defender has the ability to comprehensively obtain the basic state parameter information of the dynamic power grid when analyzing the attack behavior of the destructive false data injection attack,based on the knowledge of graph theory,a power grid dynamic model considering the false data injection is established,and a power grid dynamic model is proposed.Vulnerable variable observation and real-time attack detection method of system data transmission channel suffering from false data injection attack: First,a dynamic model of power grid is established based on the generator nodes and transmission lines in the power transmission network of the power system,considering that the data transmission channel is attacked by false data injection? Then,for the scenarios where different types of data transmission channels are attacked by false data injection,an observer-based method for observing vulnerable variables of the attacked power system is proposed.The variable credibility evaluation function of degree is used to realize real-time detection of false data injection attacks.2.Data-driven false data injection attack behavior analysis and offline attack detection methods.For the scenario where the power system defender cannot obtain some state parameter information of the system when analyzing the attack behavior of destructive false data injection attack,consider mining the characteristics of vulnerable variable time series data after being attacked by false data injection,based on the correlation relationship of the system data transmission closed loop The analysis method of false data injection attack behavior and the offline attack detection method are proposed.First,from the attacker’s point of view,the attack behavior characteristics of false data injection attack to destroy the stability of load frequency control system and the physical characteristics of false data injection attack combined with physical attack are analyzed.Then,from the perspective of the defender,the attack detection indicators for different types of attack schemes are designed considering the change characteristics of the time series data of the attacked system.An offline detection method of false data injection attack based on density-guided support vector domain description is proposed.3.Attack behavior analysis and attack detection of malicious wind power suppliers for profit in the context of electricity market.First,under the framework of the electricity market,the profit and loss of wind power suppliers in the day-ahead market and the intraday market are considered,and the motives of malicious wind power suppliers’ attack behaviors taking into account the joint contracts of wind power suppliers are analyzed.In the data transmission section,the intrusion points of false data injection attacks are discussed,and a false data injection attack model for the industrial control system of wind farms is proposed under the premise of considering the wake effect of wind farms.As well as the attack revenue in the under-power scenario,an attack optimization model aiming at maximizing the attack benefit is established from the attacker’s point of view? A detection method for false data injection attack behavior detection by network algorithm.4.False data injection attack defense multi-decision-agent game method considering the influence of game environment uncertainty.First,considering the influence of the uncertainty of the game environment on the attack-defense game matrix,a combined attack-defense game framework considering the incompleteness of the attack-defense information is proposed,and the game model is solved by using the matrix game theory? In the scenario of multiple attack(defense)decision-making individuals,an attack-defense game decision-making model is constructed under the background of the attack-defense multi-decision-making criterion,and the evidence of the attack-defense game is fused based on evidence theory,and the Nash equilibrium point in the attackdefense game is obtained through the payment matrix,which provides an analytical method for defenders to make optimal decisions in multi-body multi-criteria attack-defense games.