基于半参数回归模型的残差控制图

Residual Control Chart Based on Semiparametric Regression Model

本文探讨产品工作状态(而非工作结果)的监控,即产品质量特性值随时间和其他影响因素变化的工作过程的监控。受到卷烟生产中叶丝干燥出口物料含水率的监控问题的启发,本文提出了基于半参数回归模型的残差控制图,即通过拟合半参数回归模型获得残差,将经过标准化变换、去自相关性处理和正态变换后的残差作为描点统计量作于Shewhart控制图中,从而进行产品过程工作状态的质量监控。实例分析显示,本文方法对过程均值的异常反应灵敏,能有效地应用于实际生产中。

In this paper, we focus on monitoring products＇ working status, i.e., their longitudinal be-haviors under some factors. Inspired by monitoring the tobacco moisture at drying process in cigaretteproduction, we propose the residual control chart based on the semiparametric regression model. Throughstandardization, autocorrelation removal and normalization, the residuals are plotted on the Shewhartchart to monitor production mean process. Practical illustrations show that the proposed residual chartis sensitive to out-of-control processes.