Cascading Failure Modeling And Resilience Optimization For The Cyber-physical Power System

A Cyber-Physical Power System(CPPS),which is composed of the power(i.e.,phys-ical)and communication(i.e.,cyber)networks,is a multi-dimensional heterogeneous sys-tem that integrates communications,computations,and controls.On the one hand,the interdependencies between power and communication networks are beneficial to the es-tablishment of a safe,reliable and economical power supply system? On the other hand,such interdependencies increase the vulnerability of the system,and render it vulnerable and accident-prone,leaving countries still suffering heavy economic losses from large-scale power outages.In order to deal with the new challenges brought by the coupling effects of CPPS to the risk management of power systems,it is necessary and urgent to carry out the research on the resilience optimization of CPPS.Existing research on CPPS resilience optimization mainly focuses on the following three aspects: optimizing the re-covery sequence of damaged components,identifying and protecting critical nodes,and optimizing the coupling pattern between power and communication networks.However,there are still many challenges for current research on the optimization methods of CPPS resilience since the coupling dependencies between power and communication networks have not been well explored.Hence,this dissertation aims to investigate the cascading failure modeling methods and resilience optimization methods of CPPSs based on the coupling effects.The main research contents and contributions are summarized as fol-lows:(1)Dynamic Modeling Method of Cascading Failures for Multi-State CPPSsThe coupling effects of interdependencies between power and communication net-works and the multi-state failure problem of CPPSs have not been well investigated.For these reasons,a dynamic modeling method of cascading failures for multi-state CPPSs based on Markov process is proposed in this section,aiming to capture the dynamic changes on system performance and topology after the occurrence of disturbances and provide an universal modeling method of cascading failures for CPPSs.The proposed modeling method is the basis of the resilience research of CPPSs,and integrates the ap-proximate dynamic behavior-based and complex network-based modeling methods.The state transition between failure states is modeled as a Markov process,which involves in-formation congestion and dynamic changes of power and information flows.In addition,a robustness measure for multi-state CPPSs based on multiple degradation information is proposed,which provides a more accurate and comprehensive assessment method for the multi-state CPPS robustness.(2)Resilience Optimization Method of Multi-State CPPSs based on the Opti-mization of the Component Recovery SequenceThe multi-state failures of components,coupling dependencies within CPPSs,as well as the diversity of recovery resources and repair modes have not been investigated in the optimization method of the component recovery sequence for CPPSs.To address it,a resilience-oriented optimization method of component recovery sequence for CPPSs is proposed to minimize the reconstruction cost and the impact of disasters on system per-formance,and provide a more efficient and realistic quantitative solution to the planning challenge of component recovery sequence of CPPSs.Specifically,the above planning challenge is built as a multi-mode resource-constrained project scheduling problem,and the system resilience is quantified using the proposed time-dependent annual composite resilience metric based on the compound Poisson process.In addition,on the basis of(1),a cascading failure model of multi-state CPPSs for the recovery stage is established and embedded in the optimization model to quantify the systematic real-time performance dur-ing the recovery process and determine whether repaired components can be reconnected to the system.Finally,a modified simulated annealing algorithm considering reheating process is proposed to solve the above optimization model more accurately.(3)Resilience Optimization Method of CPPSs based on the Identification and Protection of Critical NodesThe convergence efficiency and accuracy of the critical node set identification algo-rithm in CPPSs are still low due to the lack of effective evolution and search strategies.To this end,a Gene Importance based Evolutionary Algorithm(GIEA)is proposed to iden-tify a critical node set by maximizing the total power loss,and provides a more accurate and efficient quantitative solution to the critical node set identification problem of CPPSs.Next,a node importance evaluation framework,which integrates dynamic power flows and topology information of CPPSs,is proposed to provide inputs to GIEA to improve its efficiency and accuracy.Then,the improved cascading failure model is established and applied to the evaluation framework.In addition,the concept of strong coupling node pairs based on coupling dependencies within CPPSs is proposed to reduce the dimension of the decision vector to further improve computational efficiency of GIEA.(4)Resilience Optimization Method of CPPSs based on the Optimization of the Coupling Pattern between Power and Communication NetworksThe branch rewiring and addition strategies lack quantitative programming approaches and the joint optimization strategy of these two strategies have not been investigated.For these purposes,a robustness-oriented three-layer joint optimization method for coupling patterns of CPPSs is proposed to obtain the optimal improvement strategy of coupling de-pendencies within CPPSs and the corresponding cost allocation under the limited budget.Meanwhile,the proposed optimization method provides an effective quantitative solution for branch rewiring,branch addition,and their joint optimization strategies.Specifically,the joint optimization problem of branch rewiring and addition strategies is modeled as a three-layer joint optimization model,and the improved cascading failure model in(3)is embedded into the optimization model to describe the coupling dependencies of CPPSs and quantify the system performance after the occurrence of disturbance events.Finally,the effective solving algorithms are provided for the different types of optimization prob-lems at each layer,and a three-layer hybrid solution algorithm is proposed to solve the three-layer optimization model.