Research On Malicious Attack Defense And Fault Recovery To Improve The Resilience Of Cyber-Physical Power System

Driven by the development of smart power grids,the widespread application of advanced computer technology and information technology makes the power system have gradually developed into a Cyber-Physical Power System(CPPS),with multiple coupling properties such as control,energy supply,and geography.The deep cyber-physical coupling realizes the real-time awareness of state and decision-making,which improves the operating level.The information security also further increases the vulnerable spots and leads to blackouts on occasion at the same time.The cross-space propagation caused by cyber-physcal coupling has negative effect on power system operation.To deal with this problem,the concept of resilient CPPS has been proposed.Compared with resilient power grids,the construction of resilient CPPS emphasizes the consideration of the multiple coupling relations between the information side and the phsical side,and the flexible operation methods brought by information automation technology,so as to improve the prevent ability and prepare ability for system vulnerability attacks,as well as the emergency response ability and fast recovery ability after the failure.However,the current research on attack defense for CPPS resilience improvement has deficiencies in modeling analysis for cyber-physical interaction mechanism during the fault propogation process and the cross-space transmission mechanism of the cyber-attack,and neglects the impact of information fault scenarios and component repair progress during the power system restoration.In response to these issues,through establishing a CPPS resilience evaluation framework and clarifying the improvement principles,this paper proposes a defense resources’allocation method for cyber-physical coordinated attacks,the methods of communication routing optimization and false data detection for cyber-integrity attacks,and a collaborative recovery strategy for cyber-physical compound failures.The main work and the research results are as follows:(1)The resilience quantitative metrics and assesssment framework are established from the perspective of cyber-physical coupling.The resilience quantification index and evaluation framework are proposed.The coupling models of automatic generation control system,substation automation system and energy supply relationship are established to characterize the cyber-physical interaction.An information flow model is established based on the network flow theory to simulate the operating state of the information system,and on this basis,the CPPS fault response model and the recovery model are provided.Equidispersed nonsequential Monte Carlo method is used to speed up the computation convergence.The proposed assessment method clarifies the influencing factors about resilience and provides reference for the research on improvement mearsures.(2)A defense resource allocation method is proposed based on the analysis of cross-domain cascading failure and the theory of non-zero-sum cooperative games.The model of information monitoring function is established to analyze the propagation mechanism of cascading faults under the low observation state.Then a cascading fault propagation model is provided on the basis of the interactions among phasor measurement units,circuit breakers and generators.Based on hypothesis of rational man and games of complete information,a tri-layer defense resource allocation model is established from the perspectives of defense,attack and operation to describe the offensive and defensive process.The middle and lower models are simplified through duality theory to generate the attack scheme.The upper model uses particle swarm optimization to allocate defense resources and the whole model is transformerd into a bi-level model.The proposed defense stategy can identify the vulnerable areas in the face of coordinated attacks,and provide more effective defense schemes under different attack and defense resource constraints.(3)An communication routing optimization method for the main and alternate service is proposed based on the coupling risk balance.According to the service distribution and cyber-attack risk,the cross-layer importance of imformation service is defined to characterize the relationship between information flow and physical power grids.The information entropy for link importance is established to evaluate the service transmission risk,and a multi-objective optimization model is established in consideration of the communication requirements.An improved genetic algorithm is provided to accelerate the convergence speed and realize the global distribution of the services.This optimization method can take the communication requirements and transmission risks into consideration.The risk control ability of CPPS are improved to deal with destructive cyber attacks.(4)An online defense strategy for false data injection attack(FDIA)is proposed based on historical data driven.The recurrent and dilated self-attention mechanism(RDSA)is established to expand the sampling interval of multi-head self-attention network,which is introduced into the structure of Convolutional Gated Recurrent Unit(ConvGRU).Then the ConvGRU-RDSA state prediction method is established to improve the feature extraction ability for time series.This method is combined with Kalman filter,then a hybrid state prediction method is proposed to provide supportive data for attack detection.Based on K-Nearest Neighbors and Kullback-Leibler divergence,a monitor index of hidding attack is established to relize real-time detection and correction for false data combined with consistency check and state prediction.The proposed hybrid prediction method increases the precision and stability of state estimation in normal and attack states,and the online defense strategy improves the system operation level under FDIA.(5)A cyber-physical coordinated restoration strategy for transmission system is proposed considering the maintenance program scheduling,.The time-delay models of fault localization,diagnosis and recovery beacause of communication failure are established.On the basis of that,a collaborative restoration model is proposed considering the interactions among the maintenance personnel path,information system restoration,and power system restoration.The linear process for AC power flow and clustering algorithm enhance the solve efficiency.The proposed restoration method can consider the supportting function of information system during the restoration process and the time-space coordination of equipment maintenance comprehensively,so as to speed up the progress of load recovery.