可靠性约束函数未知的网路费用最小化快速求解

针对现代网络可靠性约束函数未知的网络费用最小化问题，提出基于在线SVM和MCS技术的快速求解算法.该算法由Monte Carlo仿真方法模拟网络可靠度值，由量子粒子群算法搜寻目标全局最优解，并充分利用MCS技术模拟的可靠性信息，在线建立SVM可靠性评估模型，借助SVM模型评估后续搜寻解的可行性，减少MCS模拟次数和求解时间.与Yeh方法相比，在可靠性模拟精度为0.01的条件下，模拟次数和求解时间都缩小近1个数量级.

Fast Algorithm for Solving Cost Minimization Problem of Complex System with Unknown Constrained Reliability Function

Aiming at the cost minimization problems of modern networks with unknown constrained reliability function，a fast solving algorithm is proposed based on the Monte Carlo Simulation (MCS) technique and the Support Vector Machine (SVM) technique.The proposed algorithm uses MCS technique to obtain the reliability values of complex network，and then uses Quantum Particle Swarm Optimization (DPSO) algorithm to search for the global optimal minimal value. With the reliability information previously obtained by MCS，the online SVM evaluation model of network reliability is constructed to determine the feasibility of the subsequent solutions，reducing the MCS simulating frequencies and solving time.Experiment results verify that，compared with Yeh's method and with 0.01 reliability simulation precision，the MCS frequencies and the solving time decrease by almost one order of magnitude.