卧式加工中心现场失效数据分析实例研究

Case study on analysis of horizontal machining center field failure data

基于某大型汽车发动机厂14台卧式加工中心10个月的现场失效采集数据,首先采用卡普兰-梅尔和威布尔变换等方法,得到数据集的威布尔概率纸图;然后根据威布尔概率纸图的曲线形状,采用二重混合威布尔模型作为拟合方程;为了避免直接求解多维非线性似然方程组,将原问题转化为有约束非线性最优化问题,并采用BFGS算法和内点法求解,得到了置信度为95%的模型参数的置信区间;模型的拟合优度检验采用K-S检验方法,结论为在5%的显著性水平下接受这样的寿命分布拟合模型。得到了卧式加工中心失效率函数,为一浴盆曲线,且失效率一直下降,因此设备处于早期故障阶段。计算了卧式加工中心的平均故障间隔时间值。

Based on the field failure data of 14HMCs collected from a large automobile engine plant over ten months,the nonparametric estimation values of reliability level of HMCs were calculated by means of Kaplan-Meier method and Weibull transformation,and the Weibull Plotting Paper（WPP）plots were obtained.According to the curve shape of these WPP plots,the two-fold Weibull mixture model was applied to fit the field data set.To avoid solving the multi-dimensional non-linear likelihood equations directly,the original problem was transformed to a constrained nonlinear optimization problem which was solved by the interior point method and the BFGS formula.The confidence interval of parameter was obtained under the confidence level of 95%.The test of goodness of fit for this model was carried out by K-S test model.Result showed that such a life distribution fitting model was accepted under a significance level of 5%.Finally,the failure rate function of Horizontal Machining Center（HMC）was obtained.It was a bathtub curve with a decreasing failure rate,which meant that the HMC was running at the early stage of failure.The Mean Time Between Failures（MTBF）value of HMC was calculated to be 367hours approximately.