Enhancing multivariate process control using partial least squares: a novel strategy for managing multiple output variables

[EN] In this study, we aim to control several customer characteristics by leveraging the properties of upstream manufacturing processes. We introduce an innovative approach that combines an enhanced Partial Least Squares (PLS) regression technique, tailored to output vectors, with Hotelling&apos...

Descripción completa

Detalles Bibliográficos
Autores: Sánchez-Márquez, Rafael, Ruiz Matallana, Luis, Jabaloyes Vivas, José Manuel|||0000-0003-3411-2062
Tipo de recurso: artículo
Fecha de publicación:2025
País:España
Institución:Universitat Politècnica de València (UPV)
Repositorio:RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Idioma:inglés
OAI Identifier:oai:riunet.upv.es:10251/231405
Acceso en línea:https://riunet.upv.es/handle/10251/231405
Access Level:acceso embargado
Palabra clave:Multi-objective
Statistical process control
Multivariate
Partial least squares
Descripción
Sumario:[EN] In this study, we aim to control several customer characteristics by leveraging the properties of upstream manufacturing processes. We introduce an innovative approach that combines an enhanced Partial Least Squares (PLS) regression technique, tailored to output vectors, with Hotelling's T2 control chart. Fusing these methods allows for controlling all upstream factors influencing multiple outcomes, summarised into a single control chart for streamlined monitoring. Our method advances the current frontier of predictive Statistical Process Control (SPC), facilitating multi-objective governance of manufacturing operations. This strategy notably simplifies the task of concurrently tracking numerous end-user attributes and their corresponding upstream factors. Empirical evidence demonstrates the capability of our methodology in predicting client-specific features and detecting early signals of perturbations in upstream process stability.