| With the development of system security and reliability engineering,the concept of resilience has come into being.Looking at the relevant literature on resilience research,we have found that various fields and disciplines have different definitions of resilience,and the corresponding resilience measurement methods also have their own characteristics.The paper has further summarized the work of the predecessors and believed that the definition of resilience mainly includes two points: the influence of interference events on the system network and the network’s recovery ability after interference.On this basis,research on the resilience of the network has been carried out.At present,researches on network resilience in most literature has rarely considered damage effect of interference events on the network.The recovery research that fails to combine the damage characteristics is difficult to effectively improve network resilience.As it is unable to judge the situation of network damage,it may cause unreasonable allocation of resource and time,and the recovery strategy is researched and formulated on the basis of network damage.Therefore,the paper has conducted resilience research on the complete process of the network from the interference trigger to the execution of the recovery strategy,which has certain theoretical value and practical significance.In view of this,the paper has selected physical explosion attacks to establish the damage network model,and defined network model parameters,especially node protection coefficient and node task importance.First,the resilience recovery strategy has been established based on the quotient elasticity measurement model,and the network performance index has been measured by task importance.Further,on the basis of the classical quotient resilience measurement model,taking the parameter of recovery time into consideration,the paper has proposed an improved resilience measurement model that comprehensively considers node task importance and recovery time.Based on the improved resilience measurement model,the corresponding resilience recovery strategy has been established.In solving the problem of resilience recovery,aiming at the shortcomings of genetic algorithm and multi-chromosome genetic algorithm that the initial population is too random,easy to mature and poor solution quality,the paper has combined greedy algorithm with the genetic algorithm to propose an improved genetic algorithm which applied to resilience recovery strategy.Part of initial population of the improved genetic algorithm has been generated by greedy algorithm.Based on the existing evolution operators,the first mutation operator in the group has been designed.The improved genetic algorithm,genetic algorithm and multi-chromosome genetic algorithm have been used to solve the two recovery strategies proposed in the paper,and the solution results were compared.Finally,the paper has put forward a pros and cons index to evaluate the two recovery strategies and compared the pros and cons of recovery strategies.The simulation results show that the improved genetic algorithm running many times under the two strategies is better than the other two algorithms,and the variance of the algorithm is smaller,which shows that the improved genetic algorithm has better global solutions and strong stability.It has further proved that the network resilience is better under the recovery strategy solved by improved genetic algorithm.In addition,the two recovery strategies were compared on the using of improved genetic algorithm.The pros and cons of the recovery strategy based on the improved metric model reached 0.86,which is 16.2%higher than the recovery strategy based on the quotient model.It can be seen that the recovery efficiency of network performance under recovery strategy based on the improved measurement model is faster and the resilience value is better. |
