Abstract:  Since the failure of mechanical components was affected by the coupling effects of multiple degradation processes and external shocks and other factors, the current studies mostly considered a single factor and thought that the failure threshold was a fixed value, ignoring the facts that the equipment might reduce the shock resistance due to the degradation. Therefore, the influences of shocks on the degradation processes and the random uncertainty of the degradation processes on the shock failure threshold, as well as the correlation among different degradation processes were considered. A multi-correlation degradation reliability model was proposed considering a random variable threshold. In addition, the Copula function was used to describe the correlation among different degradation processes, and the parameter estimation method of the correlation model was given. According to the measured wheel wear data of a certain electric multiple unit(EMU), the accuracy of the model was verified by the Monte Carlo method. The influences of the correlation in the degradation processes on the reliability were discussed, and the sensitivity analysis of the random variable failure threshold was carried out to further illustrate the practicability of the proposed model.

Key words: reliability, multiple-correlation degradation, random variable failure threshold, wheel wear, degradation-threshold-shock model

摘要： 由于机械部件失效受多个退化过程及外界冲击等各种因素的耦合作用影响，而当前研究多考虑单一因素，并认为失效阈值为定值，未考虑到设备会因退化而导致其抗冲击能力降低的事实，因此同时考虑冲击对退化过程的影响，退化过程对冲击失效阈值的随机不确定性影响以及不同退化过程之间存在相关性影响，提出考虑随机变阈值的多相关退化可靠性模型。此外，通过Copula函数描述不同退化过程间的相关性，给出相关模型的参数估计方法。根据某型动车组的车轮实测磨耗数据进行实例计算，通过蒙特卡罗方法验证模型的准确性，讨论不同退化过程存在相关性对可靠度的影响，对随机变失效阈值进行灵敏度分析，进一步说明所提模型的实用性。

关键词: 可靠性, 多相关退化, 随机变失效阈值, 车轮磨耗, 退化阈值冲击模型