研究了自相关泊松计数过程单侧指数加权移动平均(exponentially weighted moving average,EWMA)控制图.基于floor和ceil取整函数,构建了单侧AF-EWMA和AC-EWMA控制图以监控泊松一阶整值自回归(INAR(1))过程,并建立二维Markov链模型计算控制图平均运行链长,以此对控制图性能进行了对比分析.计算结果表明,针对均值向上偏移,AF-EWMA图监控性能优于AC-EWMA和AR-EWMA图,同时,AF-EWMA图对控制图初始值的变动具有鲁棒性.
This paper considers one-sided exponentially weighted moving-average control chart for auto- correlated processes of count data. Based on the floor and ceil rounding function, one-sided AF-EWMA and AC-EWMA control charts are built to monitor the first-order inter-valued auto-correlated process with Poisson count data. Bivariate Markov chain model is used to calculate the average run length of these charts. On the basis of ARL values, performances of the control charts is analyzed and compared. For detecting positive shifts of the mean, it is shown that the performance of AF-EWMA chart is better than that of AC-EWMA and AR-EWMA charts. Besides, AF-EWMA chart is robust to the change of control chart starting value.