Regression on dynamic PLS structures for supervised learning of dynamic data. (August 2018)

• Record Type: Journal Article
• Authors: Dong, Yining; Qin, S. Joe
• Is Part Of: Journal of process control. Volume 68(2018)
• Page Start: 64
• Page End: 72
• Publication Date: 2018-08
• Subjects: Dynamic partial least squares -- Data-driven modeling; Process control -- Periodicals; Fabrication -- Contrôle -- Périodiques; Process control; Periodicals; Electronic journals; 660.281
• DOI: 10.1016/j.jprocont.2018.04.006
• Languages: English
• ISSNs: 0959-1524

Abstract

Highlights: A novel dynamic PLS (DiPLS) algorithm is proposed for dynamic latent variable modeling. The outer model projection is derived to ensure the latent scores are the most predictable. DiPLS gives easy-to-interpret projections with dynamic latent variables. DiPLS reduces to PLS if the latent factors have no dynamic correlations. DiPLS is efficient to model dynamic data and robust for collinear inputs. Abstract: Partial least squares (PLS) regression is widely used to capture the latent relationship between inputs and outputs in static system modeling. Several dynamic PLS algorithms have been proposed to capture the characteristics of dynamic data. However, none of these algorithms provides an explicit expression for the dynamic inner and outer models. In this paper, a dynamic inner PLS algorithm is proposed for dynamic data modeling. The proposed algorithm provides an explicit dynamic inner model that is ensured in deriving the outer model. Several examples are presented to demonstrate the effectiveness of the proposed algorithm.