Abstract: To solve the problem of zero failure intensity of Bounded Intensity Process (BIP) model at the initial time,a Modified BIP (MBIP) model was presented by introducing the location parameter into the BIP model.By taking the log-likelihood function of failure time as an optimal object,the immune clone algorithm was used to give the maximum likelihood estimators of model parameters and reliability indices.Two examples of real failure data from Numerical Control (NC) machine tools were taken to prove the proposed method,and the results of log-likelihood function and goodness-of-fit showed that the MBIP model was better than BIP model,which could provide a theoretical basis for making a reasonable maintenance strategy.

Key words: modified bounded intensity process, immune clone algorithm, maximum likelihood estimation, numerical control machine tools, reliability assessment

摘要： 针对边界强度过程模型初始故障强度为零的问题,通过引入位置参数,建立了修正的边界强度过程模型。以故障时间的对数似然函数为优化目标,应用免疫克隆算法给出了模型参数及可靠性指标的极大似然估计值。以两组数控机床现场故障数据为例对所提方法进行验证,对数似然函数值及拟合优度计算结果均表明,该模型优于边界强度过程模型,为制定合理的维修策略提供了一定的理论依据。

关键词: 边界强度过程模型, 免疫克隆算法, 极大似然估计, 数控机床, 可靠性评估