Research On Reliability Optimization Of Network Systems Based On Simulation

Reliability is an important indicator of the performance of the network system.The reliability optimization of the network can reduce the failure rate of system and improve the operation efficiency.The practical network systems always with complex structures and large scales,which lead to the difficulty to the reliability optimization.This paper focuses on the reliability optimization of network systems,which contains the methods of network system reliability evaluation,the reliability optimization model as well as the optimization algorithm.Some examples are also investigated.First of all,the simulation method to evaluate the reliability of the binary-state network is studied.Crude simulation method is inefficient for the evaluation of binary-state network reliability,and along with the large variance of reliability estimates.In this paper,the conditioned sampling plan is used to improve the efficiency of simulation,as well as the reduction of the variance.The complexity of the simulation method is also computed.To improve the accuracy in comparison of solutions,the method of common random numbers is used to generate the simulation procedure.The effect of proposed simulation method is proved theoretically,and a case of the current literature is applicated to verify the variance reduction.Then,based on the simulation method of binary-state network reliability estimation,an efficient simulation method for evaluating the multi-state network reliability is proposed.The conditioned sampling plan for multi-state network reliability evaluation is developed in this paper.A selection procedure of pathsets and cutsets is designed to reduce the sampling space.CRN is also used to reduce the variance of reliability difference of solutions.The effects of the proposed simulation method are analyzed via cases applications.The results indicate that the proposed simulation method can obtain the more accuracy reliability estimates.The third part of this paper proposes an improved genetic algorithm for network reliability optimizations.The simulation method is applied to the algorithm and the objective function is to maximize the network reliability.A ranking&selection procedure is used to allocate computation resource and improve the efficiency.In order to improved the search performance of proposed algorithm,candidate solutions are enhanced according to the constraints.Finally,a case study is analyzed to indicate the performance of proposed method.The case study includes a two-state network example and two multi-state network cases of the existing literature,and two complex network reliability optimization problems based on the problems of literature.The proposed method is applied to optimize the above problems.The results of literature problems indicate that the solutions of the proposed method are better than those in literature.The results of complex network problems denote that the improved genetic algorithm is more efficient than original genetic algorithm.In conclusion,an efficient simulation method is proposed to evaluate the reliability of network system.A mathematical model and an improved genetic algorithm of network reliability optimization is developed.Besides,examples of literatures demonstrated the performance of proposed method.