Abstract
In this article, a novel latent vector autoregressive (LaVAR) modeling algorithm with a canonical correlation analysis (CCA) objective is proposed to estimate a fully-interacting reduced-dimensional dynamic model. This algorithm is an advancement of the dynamic inner canonical correlation analysis (DiCCA) algorithm, which builds univariate latent autoregressive models that are noninteracting. The dynamic latent variable scores of the proposed algorithm are guaranteed to be orthogonal with a descending order of predictability, retaining the properties of DiCCA. Further, the LaVAR-CCA algorithm solves multiple latent variables simultaneously with a statistical interpretation of the profile likelihood. The Lorenz oscillator with noisy measurements and an application case study on an industrial dataset are used to illustrate the superiority of the proposed algorithm. The reduced-dimensional latent dynamic model has numerous potential applications for prediction, feature analysis, and diagnosis of systems with rich measurements.
Original language | English |
---|---|
Article number | e17703 |
Number of pages | 14 |
Journal | AICHE Journal |
Volume | 68 |
Issue number | 6 |
Early online date | 30 Mar 2022 |
DOIs | |
Publication status | Published - Jun 2022 |
Externally published | Yes |
Bibliographical note
Financial support for this work from a General Research Fund by RGC of Hong Kong (No. 11303421), Dimension reduction modeling methods for high dimensional dynamic data in smart manufacturing and operations, the Natural Science Foundation of China grant (U20A20189), Big data-driven abnormal situation intelligent diagnosis and self-healing control for process industries, and the City University of Hong Kong Project (9380123), SGP: Bridging between systems theory and dynamic data learning toward industrial intelligence and industry 4.0 is gratefully acknowledged. The author appreciates the helpful discussions with Prof. Alain Bensoussan at the City University of Hong Kong and Prof. Yingying Fan at the University of Southern California.Keywords
- dynamic latent variable learning
- latent system modeling
- plant-wide feature analysis
- predictable latent time series
- profile likelihood