【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。更多还原

【Abstract】 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 differential evolutionary algorithm( ADEMO / D-ENS) is proposed,which combines the decomposition with variable neighborhood 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.更多还原