Real-time Machine Learning-Based CLBF-MPC of Nonlinear Systems. Issue 2 (2020)

• Record Type: Journal Article
• Authors: Wu, Zhe; Rincon, David; Christofides, Panagiotis D.
• Is Part Of: IFAC-PapersOnLine. Volume 53:Issue 2(2020)
• Page Start: 11589
• Page End: 11594
• Publication Date: 2020
• Subjects: Machine learning -- Model predictive control -- Control Lyapunov-Barrier Functions -- Nonlinear systems -- Chemical processes; Automatic control -- Periodicals; 629.805
• DOI: 10.1016/j.ifacol.2020.12.638
• Languages: English
• ISSNs: 2405-8963

Abstract

Abstract: In this work, a real-time Control Lyapunov-Barrier Function-based model predictive control (CLBF-MPC) system using recurrent neural network (RNN) models is developed for a general class of nonlinear systems to ensure closed-loop stability and operational safety accounting for time-varying disturbances. An RNN model is first constructed for the nominal system (i.e., without disturbances) and utilized in the design of CLBF-MPC to provide state prediction. Subsequently, to improve the closed-loop performance in terms of operational safety and stability in the presence of disturbances, online learning of RNN models is incorporated within the real-time implementation of CLBF-MPC to update the RNN models using the most recent process measurement data. The proposed adaptive machine-learning-based CLBF-MPC method is evaluated using a nonlinear chemical process example.