Process structure-based recurrent neural network modeling for model predictive control of nonlinear processes

• Published in: Journal of process control 2020-05, Vol.89 (C), p.74-84
• Main Authors: Wu, Zhe, Rincon, David, Christofides, Panagiotis D.
• Format: Article
• Language: English
• Subjects: Chemical processes; Model predictive control; Nonlinear systems; Recurrent neural networks; Structural process knowledge
• ISSN: 0959-1524; 1873-2771
• DOI: 10.1016/j.jprocont.2020.03.013

•Recurrent neural network (RNN) modeling using structural process knowledge.•Economic model predictive control (EMPC) using process-aware RNN models.•Process operation and production optimization via EMPC.•Evaluation of approach using a chemical process network example. In this work, physics-based recurrent neural network (RNN) modeling approaches are proposed for a general class of nonlinear dynamic process systems to improve prediction accuracy by incorporating a priori process knowledge. Specifically, a hybrid modeling method is first introduced to integrate first-principles models and RNN models. Subsequently, a partially-connected RNN modeling method that designs the RNN structure based on a priori structural process knowledge, and a weight-constrained RNN modeling method that employs weight constraints in the optimization problem of the RNN training process are developed. The proposed physics-based RNN models are utilized in model predictive controllers and applied to a chemical process network example to demonstrate their improved approximation performance compared to the fully-connected RNN model that is developed as a black box model.