仿真基加速寿命试验优化设计方法研究

加速寿命试验可在较短时间内获得产品的可靠性信息，如何设计试验方案使加速寿命试验的结果最准确、代价最小，是加速寿命试验方案设计的一个主要问题。针对加速寿命试验解析优化方法在某些场合最优解的解析形式难以得到甚至不存在的情况，提出了一种新的基于Monte Carlo仿真的加速寿命试验优化设计方法。该方法通用性好、适用范围广。Monte Carlo方法通过大量的重复模拟试验得到问题的近似解，但要得到更精确的近似解，模拟试验的次数必须增多，因而所需要的计算量增大。利用目标函数的连续性，将非参数化曲面拟合引入优化过程，使计算量大大下降，满足了工程应用的要求。算例的结果说明了本文提出方法的正确性与有效性，敏感性分析结果表明该方法具有一定的鲁棒性。

Study of Simulation Based Optimal Designs for Accelerated Life Testing

Accelerated life testing(ALT) of a product or material is used to obtain life information quickly.It is a focus of ALT-plan designers that how to design optimal test plans for ALT.This paper presents a novelty method of Monte Carlo simulation based optimal designs for ALT.This method can be used when analytical optimization can not function.Monte Carlo method is used to obtain the approximate optimal plan by simulating test many times.More simulated tests are needed to obtain more precise solution,which may increase compute effort.In order to decrease compute effort,nonparametric surface fitting is incorporated into optimization by using continuity of object function,which meets application requirements.Results of illustrations demonstrate that this method is correct and valid.Results of sensitivity analysis show that this method is robust.