摘要
针对解析方法难以得到竞争失效产品步降加速试验最优方案和仿真法仿真规模大的难题,该文提出一种基于BP神经网络拟合的竞争失效产品步降加速试验优化方法。采用Monte-Carlo对加速试验进行模拟仿真,以最佳应力水平和试样分配比例为设计变量,以正常应力水平下各失效机理的对数特征寿命渐近方差作为目标函数,建立竞争失效产品步降加速试验优化设计模型。通过仿真实例,验证该方法有效可行。
Aiming at the problems including great difficulty of finding out the optimal plan for the step-down-stress accelerated life test of competing risk products and large simulation scale with the simulation method, the paper puts forward an optimization method for the step-down-stress accelerated life test of competing risk products based on BP neural network fitting. The method applies Monte-Carlo to have an analog simulation for the accelerated life test and establishes a model for the optimal design of step-down-stress accelerated life test of competing risk products by taking the optimal stress level and sample distribution proportion as design variables and the asymptotic variance of logarithmic characteristic life of failure mechanisms under normal stress level as objective function. The effectiveness and feasibility of the method are verified through case simulation.
出处
《中国测试》
CAS
北大核心
2016年第6期95-99,共5页
China Measurement & Test
基金
国家自然科学基金项目(61471385)
关键词
竞争失效
加速试验
优化设计
步降应力
BP神经网络
蒙特卡洛仿真
competing failure
accelerated life test
optimal design
step-down-stress
BP neural network
Monte-Carlo simulation
