Physics-informed machine learning for MPC: Application to a batch crystallization process

• Published in: Chemical engineering research & design Vol. 192; pp. 556 - 569
• Main Authors: Wu, Guoquan, Yion, Wallace Tan Gian, Dang, Khoa Le Nguyen Quang, Wu, Zhe
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.04.2023
• Subjects: Batch crystallization processes; Model predictive control; Physics-informed neural networks; Recurrent neural networks; Model predictive control; Recurrent neural networks; Batch crystallization processes; Physics-informed neural networks
• ISSN: 0263-8762
• DOI: 10.1016/j.cherd.2023.02.048

This work presents a framework for developing physics-informed recurrent neural network (PIRNN) models and PIRNN-based predictive control schemes for batch crystallization processes. The population balance model of the aspirin crystallization process is first developed to describe the formation of crystals through nucleation and growth. Then, the PIRNN modeling scheme is introduced to integrate observational data and mechanistic models for the development of machine learning models. Additionally, the physical constraints on process states are embedded in the machine learning models to prevent physically unreasonable predictions. Subsequently, the PIRNN model that captures the dynamic behavior of the batch crystallization process is utilized in the design of model predictive controller that optimizes the operation of the crystallizer. Through open-loop and closed-loop simulations, it is demonstrated that the PIRNN models using less training data achieve prediction accuracy and closed-loop performance comparable to the purely data-driven model. •Developed a physics-informed recurrent neural network (RNN) model for batch crystallization process.•Incorporated both first-principles knowledge and physical constraints into physics informed RNN models.•Investigated the performance of physics-informed RNNs with different levels of process knowledge.•Demonstrated the superiority of physics-informed RNNs over purely data-driven RNNs.