| With the close integration of information technology and modern power systems and the ubiquitous power Internet of Things,the power system has gradually evolved into a Cyber Physical System(CPS)with deep coupling of information and physical systems.Cyber physical coordinated attacks can threaten the security and stability of the power system.Quantitative analysis of the impact of cyber physical coordinated attacks on the cyber physical power system and the development of corresponding defense measures have become a key issue that needs to be solved now.Therefore,it is indispensable to establish an cyber physical coordinated attack model and solve the optimal attack strategy and risk identification of key nodes.Therefore,this paper analyzes the attack strategy of cyber physical power system.The major contributionsare as follows:Considering the problem of cross-space risk propagation of CPS,this thesis first introduces the worm propagation model into the virus diffusion modeling in power communication networks,and proposes a novel two-layer coordinated attack strategy.This strategy is aimed at the upper cyber network,and the worm propagation model is used to analyze the propagation of faults between cyber devices,which makes the model more accurate.For the power system,state estimation error caused by erroneous data injection attack(FDI)is designed to quantify the impact on the physical system.Finally,a co-simulation of the cross-space cascade fault propagation of the attacker from the cyber layer to the power system is performed on CPS system of the cyber 2PC-8 PLC node-power IEEE 14 node.The results show that the fault mechanism in the cyber layer has a great impact on the attack effect.Considering the problem of dynamic evolution of CPS under coordinated attack,this thesis uses Markov decision process to model the interaction between attackers and the CPS system to seek an optimal attack strategy.To seek the optimal strategy,this thesis designs an objective function for this optimization problem,which is to accumulate the attack benefit of the power equipment and the attack cost ratio of the cyber devices.Secondly,the Q-learning method is used to solve the optimal strategy under the model.Finally,the co-simulation is performed on the test system.The comparison results show that the coordinated attack has more damage to the CPS than the single attack method.Compared with the traditional power attack research without considering the cyber network,the weak nodes identified by the model also consider the influence of the cyber nodes and identify the high-risk nodes more accurately.To model the interaction between attackers and defenders in the CPS,this thesis proposes a dual-participant,multi-stage dynamic attack-defense game framework,where defenders and attackers utilize mixed attack and defense strategies to play this game.This framework considers the characteristics of mutual dependence and restriction between the defender and the attacker effectively.The attacker seek the optimal attack strategy under the the limited attack resources,correspondingly,the defender adopts the optimal defense resource allocation strategy against attacks.Then,the game model is solved by MinimaxQ learning algorithm,and the Nash equilibrium strategy is obtained.The action strategies of offense and defense are analyzed,which has a guiding role in the allocation of defense resources. |
