Answers are grounded in this paper and cite their sources.
Reliable monitoring of multivariate process dispersion is fundamental to sustaining quality performance and operational robustness in modern production systems. Classical control charts, however, are known to be vulnerable to outliers, particularly when parameter estimation relies on contaminated Phase I data. To address this limitation, this study develops a robust multivariate dispersion monitoring methodology for individual observations, grounded in the MEWMS and MEWMV frameworks and incorporating high‐breakdown estimators, namely ROC (Reweighted Orthogonalized Comedian) and SGK (Shrinkage Gnanadesikan Kettenring). By replacing classical estimators with their robust counterparts, the proposed approach achieves strong resistance to anomalous observations without incurring excessive computational cost. Control limits are established through comprehensive Monte Carlo simulations for multiple significance levels and smoothing parameters. The monitoring performance is systematically examined using average run length (ARL) criteria under a wide range of conditions, including dispersion shifts, varying correlation structures, and different degrees of data contamination. Comparative results demonstrate that the proposed charts consistently outperform both traditional and existing robust methods. The applicability and practical relevance of the methodology are further validated through an analysis of real chemical process data.
No takes yet. Share an insight, caveat, or question.
S et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: