测量过程控制图参数的自助最大熵融合优化

Control Parameters Optimum of Control Charts for Measurement Process Control Using Bootstrap Entropy Fusion Model

Shewhart控制图是核查数据分析的主要工具，在核查数据为小样本情况下Shewhart控制图容易导致“误发警报”。针对这一问题，提出一种自助最大熵融合方法，优化Shewhart控制图控制参数。首先，通过自助法充分挖掘核查数据自身特征，扩大样本容量。在此基础上，应用最大熵原理，描述出核查数据概率分布参数的密度函数，估计核查数据样本的均值和方差，从而优化Shewhart控制图控制参数。实验表明，经自助最大熵优化后的Shewhart控制图控制参数更加接近理论值，降低了发生“误发警报”的概率。

The Shewhart control chart is a main method to analyze check data. However, the Shewhart control chart is apt to cause false alarm under small amount of sample. For this problem, a method of Bootstrap Entropy Fusion Model is proposed to optimize control parameters of the Shewhart control chart. The bootstrap method is used to enlarge the amount of sample firstly. Based on this work, the maxim entropy is used to describe probability density functions of check data distribution parameters and estimate the mean value and variance of check data distribution. Therefore, control parameters are optimized. Experiments show that the optimized control parameters approach theory true values and the probability of false alarm is decreased.