A Weighted Correlation Coefficient Control Chart Pattern Recognition System

Abstract

Since control chart pattern recognition has the capability to recognize unnatural patterns that are some abnormal causes for variations in statistical process control (SPC), it becomes an important step in SPC. Pattern recognition techniques have been widely applied to identify unnatural patterns in control charts where most of them are capable of recognizing a single unnatural pattern for different abnormal types. However, in most real control chart applications, normal points may appear before abnormal points so that a change point from normal to abnormal may occur at any point in control charts. If we do not have a mechanism for recognizing such change patterns, we may incorrectly obtain the classification results. In Yang and Yang (2005), they created a control chart pattern recognition using a statistical correlation coefficient method. Although the control chart pattern recognition in Yang and Yang (2005) presents good properties, they had been ignored different weights in calculating the statistical correlation coefficient during different time points. In this paper, we consider a weighted correlation coefficient method to improve the control chart pattern recognition system. From comparison results, we find that our weighted correlation coefficient method actually presents better accuracy than Yang and Yang (2005).