4.8 Article

Deep Bayesian Slow Feature Extraction With Application to Industrial Inferential Modeling

Journal

IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS
Volume 19, Issue 1, Pages 40-51

Publisher

IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/TII.2021.3129888

Keywords

Feature extraction; Data models; Bayes methods; Probabilistic logic; Informatics; Linear programming; Standards; Deep learning (DL); industrial hydrocracking process; inferential modeling; probabilistic slow feature analysis (PSFA)

Ask authors/readers for more resources

Inferential modeling plays a significant role in estimating quality-related process variables in modern manufacturing. This article proposes a new nonlinear extension of probabilistic slow feature analysis (PSFA) under the deep learning framework to enhance dynamic feature extraction and improve prediction accuracy by incorporating variational inference and Monte Carlo inference. The proposed model considers the relevance of inputs with outputs to enhance prediction performance. The model is validated through an industrial hydrocracking process and achieves a significant reduction in root mean squared error compared to PSFA.
Inferential modeling has been of significance for modern manufacturing in estimating the quality-related process variables. As an effective inferential model, probabilistic slow feature analysis (PSFA) has gained attention in regression tasks to interpret dynamic properties with a slowness preference. However, PSFA is often challenged by the nonlinear sequential data due to its linear state-space structure. In this article, a new nonlinear extension of PSFA is proposed under the deep learning framework to enhance the dynamic feature extraction with limited labels, incorporating variational inference and Monte Carlo inference to derive the objective function. The proposed model considers the relevance of inputs with outputs as the input weights to upgrade prediction performance. The proposed model is verified through an industrial hydrocracking process to predict diesel yield with missing labels ranged from 0% to 50%, and the root mean squared error is reduced by at least 8.78% compared to PSFA.

Authors

I am an author on this paper
Click your name to claim this paper and add it to your profile.

Reviews

Primary Rating

4.8
Not enough ratings

Secondary Ratings

Novelty
-
Significance
-
Scientific rigor
-
Rate this paper

Recommended

No Data Available
No Data Available