基于多变量优化的恒定应力加速退化试验设计

为了实现高可靠长寿命产品加速退化试验的高效开展,提出一种可以同时优化加速应力水平、各水平下试样数、各水平测试时间等多个试验设计变量的恒定应力加速退化试验优化设计方法。基于Wiener过程与阿伦尼斯模型建立加速退化模型,以可靠寿命估计精度为目标函数,以试验总经费为约束条件,应用遗传算法搜索最优试验方案,并通过敏感性分析确定优化方案的有效性。以某碳膜电阻器为对象开展加速退化试验优化设计的实例分析来证明所提方法的合理性。敏感性分析结果表明,所提方法的优化结果具有良好的稳健性。

Constant stress accelerated degradation test design based on multivariate optimization

In order to carry out the accelerated degradation test for high reliability and long-life products efficiently, an optimal design method of constant stress accelerated degradation test are proposed, which can simultaneously optimize acceleration stress level, sample number and test time at each level. The accelerated degradation model is constructed based on the Wiener process and the Arrhenius model. The estimation accuracy of reliable life is taken as objective function, and the total cost of the test is defined as the constrain. The genetic algorithm is applied to realize the search of the optimal test scheme, and the effectiveness of the optimal scheme is rerified through the sensitivity analysis. An accelerated degradation test for a carbon film resistor is optimally designed as a case study to demonstrate the rationality of the proposed method. Finally, the sensitivity of the model parameters to the optimization results is analyzed to illustrate the robustness of the optimal design procedure.