新的MOPSO及其在大型复杂系统可靠性优化中的应用  

Novel MOPSO and its application in reliability optimization of large complex systems

作　　者：鲁鹏[1] 章卫国[1] 李广文[1] 刘小雄[1] 

机构地区：[1]西北工业大学自动化学院,西安710072

出　　处：《计算机应用研究》2012年第6期2153-2155,2160,共4页Application Research of Computers

基　　金：航空科学基金资助项目(20090753008)

摘　　要：可靠性优化问题是大型复杂系统设计的一个关键问题。针对大型复杂系统多个指标(可靠度、造价和冗余数)同时进行最优分配的结果多样性不好的问题,提出了一种基于杂草克隆的多目标粒子群算法—IWMOP-SO(invasive weed multi-objective particle swarm optimization)的多指标分配方法。该分配方法通过引入杂草克隆机制来改善Pareto最优解的收敛性和多样性。通过对大型复杂系统多个指标进行分配,其分配效果与NSGA-Ⅱ相比,得到的Pareto非劣解集多样性和均匀性好,分布范围更广,更利于设计者进行决策,是一种更有效的复杂系统多指标分配方法。Reliability optimization problem is a key problem of designing the large complex system.As there existed a diversity problem in the multi-indexes（such as reliability,cost and redundancy） allocation optimization,this paper proposed a novel multi-indexes allocation algorithm-IWMOPSO based on the multi-objective optimization.The algorithm introduced invasive weed mechanism in order to better the convergence and the diversity of the Pareto front.It utilized the algorithm to allocate the multi index of the large complex systems.In order to indicate its effectiveness and efficiency,the achieved Pareto front was compared with the result achieved by NSGA-Ⅱ.It turns out that the new algorithm gets a better diversity and convergence,and it is more convenient for designers to make a decision,therefore,it is a more effective multi-objective allocation algorithm of large complex systems.

关 键 词：可靠性优化 多指标分配 杂草克隆 多目标粒子群算法 最优分配 多样性 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]