Cascading Failure Mechanism And Mitigation Strategies Of Smart Grids Under Attacks

Smart grid integrates power system engineering with information and communication technology to form a complex heterogeneous system.As control decisions,contingency analyses,and protection operations are deeply dependent on the reliable and secure communication network,the smart grid is vulnerable to cyber attacks.False data injection attack is a typical cyber-physical attack threatening the smart grid security.It can monitor,infiltrate,and falsify transmission data to attack the communication network,then affect decision-making and send wrong control instructions to cause misoperations or delayed protections of physical equipment,triggering faults transmitted interactively between networks and leading to large-scale cascading failures even blackouts.Based on the complex network theory,it is necessary to study the cascading failure propagation caused by this attack so as to identify the general vulnerability of the power system topology.Combining the mechanism of power flow distribution with the principle of communication information transmission,the evaluation indexes are used to assess the influences and finally help to form mitigation strategies.Load redistribution based sequential attack model of smart grids is used to analyze the topology vulnerability.The node types and their location importance can be used to identify the most vulnerable node.The attack model integrates the operation characteristics of the power system with the giant component principle of the percolation theory,in order to analyze the general vulnerability of the power system topology.From the analysis of the experimental results,we find that sequential attacks are more destructive than simultaneous attacks,and the topology vulnerability is closely related to the critical node types and locations.Based on the complex network theory,a smart grid capacity model is proposed according to the power system operation mechanism and the communication network information routing principle.Power flow distribution obeys Kirchhoff's law and conservation of energy,while communication information transmission follows routing protocol policy.Based on the load-capacity model,a hybrid capacity model with a unified tolerant parameter is therefore proposed to integrate the power flow capacity with the communication transmission capacity.Simulation results show that when the tolerance coefficient is greater than 2,the cascading failure propagation process tends to be relatively smooth.Based on the smart grid system capacity model,influence factors of cascading failure propagation are analyzed to construct the evaluation index system of cascading failure robustness.In the node-attack cascading failure model,the cascading failure propagation is a Markov process.Considering the interdependent degree similarity,system capacity,attack type,attack orientation and attack range,evaluation indexes for islanding operations are employed to analyze the influence on the cascading failure propagation.These indexes include network connectivity level,system loss degree and power flow shift rate.It is thus useful to guide cascading failure mitigation strategies of smart grids.From the experimental results,these influence factors have various effects on the robustness of the interdependent networks and validate the effectiveness of the index system.According to the engineering demands,it is necessary to adjust the original network topology in order to alleviate the cascading failure impacts.Combined with the islanding characteristics and the node vulnerability,three low-degree-node based link-addition strategies are proposed based on the community partition principle.The remaining island ratio and the sequential attack difficulty coefficient are proposed to assess the robustness of smart grids.From the analysis of the experimental results,the three proposed strategies enhance the network connectivity,optimize the network planning and improve the robustness of both single networks and interdependent networks under three attack scenarios.