Abstract: Aiming at the circumstance of stochastic process mean shift,a statistical-economic design method of EWMA control chart was proposed.In this method,the three phases of the in-control condition,the out-of-control condition without signal and the maintenance condition to restore were defined as a cycle to analyze the average time and the quality cost structure of the three phases.The quality losses caused by the defective products were quantified by computing the probability that the product quality characteristic fell beyond the specification limit.The optimal design model of EWMA control chart was set up to minimize the expected cost per time unit,and the genetic algorithm was designed to obtain the optimal EWMA control chart parameters such as the sample size,the sampling interval,the smoothing factor and the control limit of the chart.A number of comparative experiments were conducted to show that the proposed model was superior to the Shewhart X control chart and the traditional EWMA control chart.

Key words: statistical process control, exponentially weighted moving average, optimization model, quality loss, process mean shift

摘要： 针对过程均值偏移随机的情况,提出一种统计经济最优的指数加权移动平均控制图优化设计方法。该方法将过程受控、失控未检出、失控被检出并进行恢复这三个阶段定义为一个周期,分析了三个阶段的平均时长及质量成本构成,通过计算产品质量特性超出规格界限的概率量化缺陷产品所造成的质量损失,以单位时间内期望成本最小为目标建立指数加权移动平均控制图优化模型并设计了遗传算法,优化了样本容量、采样间隔、平滑界限和控制界限等参数。通过与休哈特均值控制图、传统指数加权移动平均控制图等进行对比验证了该模型的优越性。

关键词: 统计过程控制, 指数加权移动平均控制图, 优化模型, 质量损失, 过程均值偏移