Research On Assessment And Recovery Strategy Of CPPS Considering Cascading Failure

With the continuous construction of the national smart grid and the continuous transformation and upgrading of the power system,the traditional power system has gradually developed into the Cyber-Physical Power System(CPPS)cross-applied by information technology and physical technology.Such converged systems offer significant advantages in terms of improved energy efficiency,optimal resource allocation and reduced operating costs,yet they also present new security challenges.Security problems in any one link may lead to system collapse,power outage and other serious consequences,so it becomes especially important to guarantee the safe and reliable operation of the system.Based on the above background,this paper investigates the problems of the cascading failure modeling,node vulnerability assessment and sequential recovery decision for the CPPS.The main contributions include the following three aspects.(1)Considering the power flow characteristics,data transmission and the economic dispatch,the modeling process is divided into three parts: power system,cyber system and communication system.The DC power flow is used to calculate the power system information,and the optimal dispatch model of the cyber system is established with the objective of minimizing generating cost.The power system and the cyber system are effectively related through the uplink/downlink communication channel model,so as to construct an integrated mathematical model of the CPPS.Based on the unified model,the propagation mechanisms of three typical cyber-attacks(measurement and control equipment attack,denial of service attack and false data injection attack)and their formulaic expressions are studied.Meanwhile,the cascading failure process of the CPPS under different cyber-attacks is analyzed.(2)To address the problems of one-sided assessment indexes and inaccurate index weights of existing node assessment methods,a hierarchical node vulnerability index system is constructed from both the power flow characteristics of the power grid and the system topology security.The improved fuzzy analytic hierarchy process is used to assess the node vulnerability from qualitative and quantitative perspectives.The vulnerability evaluation results can provide a valid reference for system operation and maintenance,and promote the improvement of robustness.(3)It is equally important to study the rapid recovery of CPPS from cascading failures to normal operation and to study the fault propagation mechanism of cascading failures.From the defender’s perspective,the sequential recovery model is built within the framework of the cascading failure model with the goal of maximizing load recovery.For different scale systems,the practical sequence recovery strategies based on Q-learning and deep Q-network algorithm in reinforcement learning are developed to effectively determine the best recovery solution for the failed components.The effectiveness and superiority of the proposed strategy are also verified on various test cases considering specific scenarios such as cyber-attacks and distributed generation access.