Statistical design of individual and integrated multivariate hotelling T 2 control chart for a multi-component load-sharing system subject to dependent failure modes
Year: 2026
Type: article
Abstract: In this paper, with the aim of a novel comparative study, we propose individual and integrated multivariate Hotelling T2 control charts for a multi-component load-sharing system. In fact, the multivariate control chart for monitoring process under assumptions that the vector of system signals follows a multivariate normal distribution is considered. To represent a real-world multi-component system, it has been assumed the system is subject to both hard and soft failures which are mutually dependent and the severer soft failures would entail the rise in the probability of the occurrence of hard failures. The degradation states are governed by a discrete time Markov chain (DTMC) and moreover, to model the proposed dependent hard and soft failures, a proportional intensity model (PIM) is applied. The objective is to compare individual Hotelling T2 control chart performance with an integrated multivariate control chart considering inherently correlated signals of the system components by computing average run length (ARL) and standard deviation run length (SDRL). Sensitivity analysis is conducted to evaluate the effect of load-sharing parameters on the ARL. Computational results show that the proposed control chart is sensitive to model parameters and the superiority of the integrated control chart depends on the load sharing conditions which affect the correlation structure of the component signals.
Institution: K. N. Toosi University of Technology
Language: English
Subfield: Statistics, Probability and Uncertainty
Field: Decision Sciences
Domain: Social Sciences
Open Access status: green
OpenAlex ID: W4319457677