加速退化模型及外推结果准确度的定量验证方法

加速退化试验本质上是牺牲可靠度评估精度换取可靠度评估效率，外推出的可靠度结果通常会与真实值存在一定的偏差，因此，验证加速退化模型与外推结果的准确度是非常必要的。根据可靠性建模的步骤设计了验证加速退化模型及外推结果的技术流程，结合Wiener-Arrhenius加速退化模型构建了验证技术框架。采用了基于假设检验的模型验证方法，利用Kolmogorov-Smirnov检验法、Anderson-Darling检验法分别验证Wiener退化模型、加速退化模型及外推的可靠度模型是否合理。提出了具有工程实用性的基于面积比的外推结果验证方法，利用蒙特卡罗仿真解决复杂可靠度函数积分问题，用于定量表征外推结果的准确度，通过面积比阈值判定是否接受外推结果。通过惯导系统伺服电路实例应用说明了技术框架的可行性与有效性，研究工作为解决加速退化试验中的验证难题做出了有益的探索，所提出的各种定量验证方法具有较好的工程应用价值。

Quantitative validation methods for accelerated degradation model and extrapolated results

The accelerated degradation test improves the efficiency of reliability assessment by sacrificing some assessment accuracy, and there are commonly deviations between extrapolated reliability results and true values. Thus, it is necessary to validate the accuracy of the accelerated degradation model and extrapolated results. The technical flow for validating the accelerated degradation model and extrapolated results are designed according to the steps of reliability modeling, and a validation technology framework is constructed with the Wiener-Arrhenius accelerated degradation model. The model validation method based on hypothesis test is adopted. The application reasonability of the Wiener degradation model, accelerated degradation model and extrapolated reliability model is validated by the Kolmogorov-Smirnov and Anderson-Darling tests. A practical method for validating extrapolated results based on the area ratio is proposed, which solves the integral problem of complex functions using Monte Carlo simulation. The area ratio is applied to quantitatively describe the accuracy of extrapolated results, and a threshold of area ratio is specified to determine whether to accept the extrapolated results or not. The feasibility and effectiveness of the proposed technology framework is demonstrated by a case study of the servo circuit of the inertial navigation system. The proposed quantitative validation methods show great applicability in practice.