针对解析方法难以得到或者不能得到步降加速退化试验最优方案的难题,提出一种基于粒子群算法的Monte-Carlo仿真步降应力加速退化试验优化算法.该算法通过大量的重复模拟试验生成试验退化数据,寻找最佳监测频率、检测次数和样本量,以正常使用应力下的对数p阶分位寿命渐近方差估计最小为目标,采用粒子群算法对退化试验数据采用极大似然估计进行统计分析,建立了基于仿真的步降应力加速退化试验优化设计模型.基于算例,给出了不同约束条件下的优化设计方案,得到了该方法也满足小子样产品步降加速退化试验优化设计的结论,并最终得到其最优试验方案.

For the problem that an optimal scheme for step-down-stress accelerated degradation test is difficult to or cannot be obtained through analytical methods,an optimization algorithm of Monte-Carlo simulation step-down-stress accelerated degradation test (SDSADT) based on Particle Swarm Optimization (PSO) is presented.This algorithm generates test degradation data through a number of repeated simulation tests,and finds the optimal monitoring frequency,detection times and sample size.With the minimum logarithmic asymptotic variance estimation of 100pth percentile of the lifetime distribution of the product at use condition as the objective,a statistical analysis of the degradation test data is made by using Maximum Likelihood Estimation (MLE) theory based on PSO,and a model of optimization design for simulation based SDSADT is established. Based on an example,the optimal design scheme is given under different constraint conditions,with the conclusion that this method is also applicable to optimization design for SDSADT of small-subsample product,and the final optimal test scheme is obtained.