Monitoring the process quality for multistage systems with multiple characteristics

Purpose: The purpose of this paper is to provide a new approach for detecting the small sustained process shifts in multistage systems with correlated multiple quality characteristics. Design/methodology/approach: The authors propose a new multivariate linear regression model for a multistage manufacturing system with multivariate quality characteristics in which both the auto-correlated process outputs and the correlations occurring between neighboring stages are considered. Then, the multistage multivariate residual control charts are constructed to monitor the overall process quality of multistage systems with multiple quality characteristics. Moreover, an overall run length concept is adopted to evaluate the performances of the authors’ proposed control charts. Findings: In the numerical example with cascade data, the authors show that the detecting abilities of the proposed multistage residual MEWMA and MCUSUM control charts outperform those of Phase II MEWMA and MCUSUM control charts. It further demonstrates the usefulness of the authors’ proposed control charts in the Phase II monitoring. Practical implications: The research results of this paper can be applied to any multistage manufacturing or service system with multivariate quality characteristics. This new approach provides quality practitioners a better decision making tool for detecting the small sustained process shifts in multistage systems. Originality/value: Once the multistage multivariate residual control charts are constructed, one can employ them in monitoring and controlling the process quality of multistage systems with multiple characteristics. This approach can lead to the direction of continuous improvement for any product or service within a company.