A Study of Non-Normal Process Capability Analysis Based on Box-Cox Transformation

Abstract

Process capability indices are the important tools used in most of the manufacturing industries to check whether the manufactured products meet their quality specifications or not. Process capability analysis requires that the quality characteristic data be normally distributed. In actual production, a lot of stable processes do not necessarily satisfy the assumption of normal distribution. An approach to tackle this problem is to use the appropriate transformation methods to convert these non-normal data. Therefore, a method of converting non-normal data into normal data is proposed so that the data can be analyzed using the process capability indices. In this paper, an improved Box-Cox transformation model is proposed to deal with non-normal data and calculate its process capability indices, and the concrete steps are given. Finally, the method is used to study the actual cases, and the process capability indices are calculated. The effectiveness and practicability of the method are proved by comparison with the actual situation. In this paper, Minitab analysis software is used to assist the realization of this method. It has strong operability and convenience, and can be used to guide production practice.