Although there are various methods of improving the quality of products, the statistical process control (SPC) is regarded as the most scientific and valid approach. Since the
¯ X-chart uses only
current observations, it is not sensitive to small means shift. To overcome this weakness, the cumulative sum (CUSUM) control chart [1] and the exponentially weighted moving average (EWMA) control chart [2] have been developed. Since these charts use not only the current observations but also historical information, they are more sensitive to small changes than the
¯ X-chart. Hunter [3]
argued that the EWMA facilitates the acquisition of control limits and allows for an empirical dynamic control equation. Lucas and Saccucci [4] showed that the performance of the EWMA chart is similar to the CUSUM chart.
In most industries, quality variables to be monitored are not just single but multiple. In this situation, the univariate control chart can be applied to each of quality variables separately. However, the univariate control chart may not detect the process shift if the variables are correlated with each other. To overcome this difficulty, multivariate control charts have been proposed based on chi-squared statistics and Hotelling’s T-squared statistics [5]