Monitoring multivariate profiles in multistage processes

Abstract

In some quality control applications, processes consist of multiple components, stations or stages to finish the final products or the services. Some quality characteristics in each stage of these processes (called multistage processes) can be represented by a relationship between a response and one or more explanatory variables which is named as profile. In this paper, a general model is proposed for monitoring multivariate profiles in multistage processes. To this aim, the multivariate form of the U transformation approach is first used to remove the effect of the cascade property between the stages. Then, three control schemes are employed to monitor the parameters of multivariate simple linear profiles in multistage processes in Phase II. The performances of the proposed schemes are evaluated in terms of the average run length criterion using extensive simulation experiments. The results show that the transformation approach efficiently eliminated the effect of the cascade property involved in multistage processes and that all of the schemes perform well under different conditions. In the end, an illustrative example is given to demonstrate the application of the proposed methodology.

Keywords:

• Multivariate profile
• Multistage process
• Cascade property
• Autocorrelation
• Statistical quality control

Acknowledgment

The authors are thankful for constructive comments of the anonymous reviewers. Taking care of the comments significantly improved the presentation of the manuscript.