Abstract: Aiming at the long average alarm time of Exponentially Weighted Moving Average (EWMA) t control chart under uncontrolled state,a Variable Sampling Interval Adaptive EWMA t (VSI-AEWMA t) control chart was proposed.The smoothing parameter was adjusted adaptively to improve sensitivity of EWMA t control chart for detecting large size of assignable cause,and the average time to signal of EWMA t control chart was shortened by variable sampling interval algorithm.Markov chain method was used to calculate the average time to signal of VSI-AEWMA t control chart,and the corresponding parameter optimization algorithm was given.VSI-AEWMA t control chart was compared with VSI-EWMA t control chart with simulation analysis,and the result showed the better performs of VSI-AEWMA t control chart.

Key words: adaptive EWMA, t control chart, variable sampling interval, average time to signa, ;multi-objective optimization

摘要： 针对平滑系数固定的EWMA t控制图失控状态下平均报警时间长的问题,提出一种变采样间隔、自适应平滑系数的EWMA t(VSI-AEWMA t)控制图。平滑系数的自适应特性加强了EWMA t控制图对大偏移的检测能力,变采样间隔特性缩短了EWMA t控制图的平均报警时间。采用Markov链方法计算VSI-AEWMA t控制图的平均报警时间,给出了VSI-AEWMA t控制图的参数优化算法。通过仿真分析比较了VSI-AEWMA t控制图与VSI-EWMA t控制图的统计性能,结果表明VSI-AEWMA t控制图的性能优于VSI-EWMA t控制图。

关键词: 自适应EWMA, t控制图, 变采样间隔, 平均报警时间, 多目标优化