Shapley-Based Multi-Objective Distributional Control of Cascaded Refining Systems With Deep Temporal Modeling
IEEE Transactions on Automation Science and Engineering · 2026 · 23 · 6132-6146
DOI: 10.1109/tase.2026.3672526 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Automation Science and Engineering, 2026. Verfasst von Mingjie Zheng und 1 weiteren Autor:innen. Die Arbeit nennt 36 Literaturangaben. Gefördert von: National Natural Science Foundation (NNSF) of China, Guangdong Basic and Applied Basic Research Foundation, Guangdong Generic Institution Innovation Team Research Foundation.
Abstract (Originalsprache)
Accurately controlling cascaded refining processes is usually difficult in pulp and papermaking systems when using model predictive control (MPC), as the MPC framework requires the simultaneous regulation of deterministic targets and the reshaping output probability density functions under nonlinear and large time-delay conditions. Deep learning can facilitate advanced data-driven modeling and can be assimilated into MPC, but is usually limited by the model interpretability and the complex requirements of multi-objective distributional control. This study proposes a deep neural network-based MPC framework that integrates a predictive model with interpretable optimization strategies to overcome the aforementioned limitations. To enable precise prediction, we first develop a temporal autoencoder network that simultaneously forecasts the Schopper-Riegler beating degree and parameterizes the fiber length distribution, as well as to effectively capture the system’s nonlinear characteristics. Furthermore, to tackle multi-objective challenges, we propose an optimization strategy that integrates a predictive cost-driven adaptive weighting mechanism and an adaptive gradient descent algorithm to dynamically balance objectives and accelerate convergence to optimal control. Also, to address interpretability and optimal control issues, we formulate a dynamic Shapley value analysis to adaptively refine the control optimization feasible region by capturing time-varying feature interactions. Finally, the efficiency of the proposed model is validated through a case study on this system, demonstrating that significant improvements can be achieved in terms of control performance, response accuracy, and interpretability. Note to Practitioners—This work considers high-accuracy control of the cascaded refining process in the pulp and paper industry. It is critical to regulate both the Schopper-Riegler beating degree and the fiber length distribution by capturing long-term dependencies and dynamic process features from multivariate time series data. Yet, this is an extremely challenging task because the refining process is a complex multistage system with strong time delays, nonlinear transitions, input hysteresis, and unobservable intermediate states. These issues make it difficult to maintain stable pulp quality and hinder the automation of industrial refining operations. Current control methods are typically based on mechanism modeling or standard data-driven approaches. However, they often suffer from poor adaptability, low interpretability, and limited accuracy when dealing with multiple objectives and complex process dynamics. As a result, they fail to provide reliable quality control under real production conditions. This work proposes an interpretable predictive control framework that integrates a temporal autoencoder network with model predictive control. The method simultaneously forecasts both scalar and distributional pulp quality indicators, and includes a predictive cost-guided weighting mechanism to dynamically balance competing control objectives. A gradient-based optimization method ensures fast convergence, and a dynamic Shapley value analysis is used to enhance interpretability by capturing time-varying feature interactions. Experimental results on real cascaded refining data show that the proposed method improves control precision, robustness, and transparency. In the future, this method can be applied to other nonlinear industrial process control problems, such as multistage manufacturing and process optimization. It may also provide methodological insights for optimization-based control and decision-making frameworks that integrate learning-based predictive models in autonomous systems.
Förderung
- National Natural Science Foundation (NNSF) of China
- Guangdong Basic and Applied Basic Research Foundation
- Guangdong Generic Institution Innovation Team Research Foundation
- Advanced Papermaking Joint Laboratory Open Project
Zitieren
APA
Zheng, M., & Liu, Y. (2026). Shapley-Based Multi-Objective Distributional Control of Cascaded Refining Systems With Deep Temporal Modeling. IEEE Transactions on Automation Science and Engineering, 23, 6132-6146. https://doi.org/10.1109/tase.2026.3672526
Vancouver
Zheng M, Liu Y. Shapley-Based Multi-Objective Distributional Control of Cascaded Refining Systems With Deep Temporal Modeling. IEEE Transactions on Automation Science and Engineering. 2026;23:6132-6146. doi:10.1109/tase.2026.3672526
BibTeX
@article{Zheng2026,
author = {Zheng, Mingjie and Liu, Yiqi},
title = {Shapley-Based Multi-Objective Distributional Control of Cascaded Refining Systems With Deep Temporal Modeling},
journal = {IEEE Transactions on Automation Science and Engineering},
year = {2026},
volume = {23},
pages = {6132--6146},
doi = {10.1109/tase.2026.3672526}
}
EUVIMED