ABSTRACT Most existing multivariate control charts possess directional invariance, which means that they are designed to detect shifts in the mean vector of the process in all directions. In many applications, quality managers are often primarily concerned with abnormal signals in a particular direction. In such cases, the lack of directional sensitivity represents a key limitation of traditional multivariate control charts. To address this deficiency, we develop a set of new directional multivariate exponentially weighted moving average (MEWMA) charts with false discovery rate (FDR) control and an adaptive sampling scheme, specifically the variable sampling interval (VSI) strategy, in comparison with the conventional fixed sampling interval (FSI) strategy to monitor process means, mitigating sequential error build‐up, and optimizing sampling efficiency simultaneously. The paper provides a detailed comparison of the proposed control charts with competitors through extensive numerical experiments, offering valuable information to practitioners. The results show that the proposed charts typically outperform existing one‐sided and two one‐sided MEWMA control charts while effectively controlling the FDR. Furthermore, case studies are presented using real datasets, where the proposed control charts are applied innovatively to a log‐binomial model.
Feng et al. (Tue,) studied this question.