Abstract: To reduce the uncertainty of Non-Homogeneous Poisson Process (NHPP) model and improve the assessment accuracy of NC machine tools,Bayesian reliability analysis results which included point and interval estimations of model parameters and reliability indices were given with Markov chain Monte Carlo method.Two real engineering examples were presented based on the maintenance experiences of field experts,the reliability of NC machine tools with increasing and decreasing failure intensity in minimal repair were analyzed respectively.The results showed that the length of interval estimate by using Bayesian method was narrower than that obtained by Maximum Likelihood Estimation (MLE),and reliability analysis results of the former was more close to the actual cases.

Key words: minimal repair, small-sample data, Bayesian reliability, numerical control machine tool, Markov chain Monte Carlo

摘要： 为减少非齐次泊松过程模型的不确定性,提高数控机床的评估精度,利用马尔科夫链蒙特卡洛法给出了数控机床最小维修的贝叶斯可靠性分析结果,包括模型参数和可靠性指标的点估计和区间估计。通过两个工程实例,结合专家现场维修经验,分别分析了最小维修数控机床故障强度减小和增大时的可靠性,计算结果显示贝叶斯可靠性分析的区间估计长度小于极大似然估计,其分析结果更加全面和接近实际情况。

关键词: 最小维修, 少样本数据, 贝叶斯可靠性, 数控机床, 马尔科夫链蒙特卡洛法