Grid Vulnerability In Cyber Attack Scenarios

Nowadays,our work and life are highly dependent on many infrastructure networks,such as power networks,communication networks,and transportation networks.However,there are many security issues in these infrastructure networks.In the power network,the widespread application of information technology has led to a continuous increase in the risk of network attacks,which is likely to further cause cascading failures and large-scale power outages.Therefore,it is of great practical significance to understand and optimize the vulnerability of power networks.Based on the complex network analysis method,we model and analyze the attack behaviors and cascading failure processes in the scenario of power grid,and use intelligent optimization algorithms to optimize the power business flow and power grid topology to improve the robustness of the power grid.Our work helps to design and plan the power networks.The main research contents of this thesis are as follows:1.We studied how the configuration of the importance of the power business affects the vulnerability of the power communication network under attacks.Specifically,to reduce the vulnerability of the power communication network,we propose an adaptive importance configuration strategy of business flow based on node distance.Simulation experiments show that under the condition of intention link attack,the strategy has the best configuration parameters leading to the least vulnerability of the power communication network.Moreover,the simulated annealing algorithm is used to optimize the transmission path between the node pairs to improve the robustness of the power communication network.2.We studied the problem of cascading failures in power grids under the condition of link attacks.We proposed a link centrality metric based on network topology and electrical characteristics,and further provided a greedy attack algorithm and an optimal attack algorithm based on the link centrality.Simulation experiments were conducted on the IEEE standard data set.The experimental results show that the scale of the cascading failure caused by the optimal attack algorithm is larger than the greedy attack algorithm and the classic PSO algorithm.The computational cost of the greedy attack is lower than that of the optimal attack and PSO-based attack.In addition,we analyzed the impact of the number of attacked links and the number of iterations of the particle swarm on cascading failures.3.we studied the optimization of the interconnections between the interdependent cyber and power networks.We used the scale-free network models and random network models to generate interdependent network structures.Considering the fragility of power dependent networks under cascading failures and the cross-network transmission efficiency of power control signals,we constructed a dual-objective optimization problem,and adopted the MOPSO-CD algorithm and MOEA/D algorithm to solve this problem.Furthermore,we employ the node degree and clustering coefficient to improve the MOEA/D algorithm.Experimental results show that,compared with the MOPSO-CD algorithm and the MOEA/D algorithm,the improved MOEA/D algorithm can get a better Pareto frontier.