基于ADEMO/D-ENS算法的表决冗余系统多目标可靠度优化与分配  

Multi-Objective Reliability Optimization and Allocation of Redundancy System with Voting based on ADEMO / D-ENS Algorithm

作　　者：刘志君[1] 高亚奎[2] 章卫国[1] 胡云池[3] 

机构地区：[1]西北工业大学自动化学院,陕西西安710129 [2]中国航空工业集团公司西安飞机设计研究所,陕西阎良710000 [3]空军航空大学军事教育训练系,辽宁长春130022

出　　处：《西北工业大学学报》2014年第6期987-993,共7页Journal of Northwestern Polytechnical University

基　　金：航空科学基金(20125853035)资助

摘　　要：为了解决具有表决冗余结构的系统可靠度多目标优化与分配问题,提出了一种新颖的算法——基于解耦的邻域种群集多目标自适应差分进化算法(ADEMO/D-ENS),该算法将变邻域解耦算法和PM自适应差分进化算法相结合,不仅克服了经典差分进化算法存在的缺陷,同时解决了解耦算法邻域种群选择的问题,并将该算法和NSGA-2算法进行了仿真对比。最后,将ADEMO/D-ENS算法用于某系统可靠度和成本的多目标优化,确定了系统可靠度和成本的Pareto前沿,并给出了决策者感兴趣的组件可靠度和冗余度数据。In order to solve optimization and allocation of multi-objective reliability redundancy system with voting, a novel algorithm based on decomposition with ensemble of neighborhood sizes and multi-objective adaptive differen-tial evolutionary algorithm （ ADEMO/D-ENS） is proposed, which combines the decomposition with variable neigh-borhood sizes and probability matching adaptive differential algorithm.The proposed algorithm not only overcomes drawbacks of the classical differential evolutionary method, but also solves selection of neighborhood size based on decomposition technique.Simulation contrast between ADEMO/D-ENS algorithm and NSGA-2 was done.Finally, the proposed algorithm was employed to do multi-objective reliability and cost optimization of a certain system, and the pareto front between system reliability and cost is obtained.Meanwhile, component reliability and redundancies, which the decision maker may be interested in, were given.

关 键 词：可靠度 多目标优化 自适应算法 冗余 PM自适应 差分进化 

分 类 号：V249.4[航空宇航科学与技术—飞行器设计]