(92c) Learning to Accelerate Process Optimization Via Artificial Intelligence

Decision-making (optimization) problems arise in a wide range of applications in chemical engineering. Typical examples include process design, production planning and scheduling, and process control [1,2]. Although efficient numerical algorithms have been developed for the solution of a wide class of optimization problems [4-6], the online solution of large-scale and mixed integer optimization problems remains challenging due to the limited computational budget. Recently, machine learning (ML) has been used to accelerate the solution of optimization problems [7-12]. This approach leverages data to learn the behavior or the output of the optimization algorithm for a given problem based on some appropriate feature representation of the problem.

In this talk, we present a data-driven framework for 1) learning to select the best solution strategy and 2) tuning it. The approach relies on machine learning to 1) build a classifier that predicts be best solution strategy and 2) provide good initialization for cutting-plane-based algorithms. The framework is applied to an economic mixed integer model predictive control problem, which arises in the operation of multi product production processes. The results show that this approach can guide the selection of the best solution strategy, thus reducing the solution time, without compromising solution quality.

References:

[1] P. Daoutidis, J. H. Lee, I. Harjunkoski, S. Skogestad, M. Baldea, C. Georgakis, Integrating operations and control: A perspective and roadmap for future research, Computers & Chemical Engineering 115 (2018) 179–184.

[2] E. N. Pistikopoulos, A. Barbosa-Povoa, J. H. Lee, R. Misener, A. Mitsos, G. V. Reklaitis, V. Venkatasubramanian, F. You, R. Gani, Process systems engineering–the generation next?, Computers & Chemical Engineering 147 (2021) 107252.

[3]. Biegler, L.T., 2024. Multi-level optimization strategies for large-scale nonlinear process systems. Computers & Chemical Engineering, 185 (2024) p.108657.

[4]. M. Tawarmalani, N. V. Sahinidis, A polyhedral branch-and-cut approach to global optimization, Mathematical programming 103 (2) (2005) 225–249.

[5] F. Boukouvala, R. Misener, C. A. Floudas, Global optimization advances in mixed-integer nonlinear programming, minlp, and constrained derivative-free optimization, CDFO, European Journal of Operational Research 252 (3) (2016) 701–727.

[6]. Bengio, Y., Lodi, A. and Prouvost, A., Machine learning for combinatorial optimization: a methodological tour d’horizon. European Journal of Operational Research, 290(2) (2021), pp.405-421.

[7]. Ren, Y.M., Alhajeri, M.S., Luo, J., Chen, S., Abdullah, F., Wu, Z. and Christofides, P.D., A tutorial review of neural network modeling approaches for model predictive control. Computers & Chemical Engineering, 165 (2022), p.107956.

[8]. Kumar, P. and Rawlings, J.B., Structured nonlinear process modeling using neural networks and application to economic optimization. Computers & Chemical Engineering, 177, (2023), p.108314.

[9]. Mitrai, I. and Daoutidis, P., Taking the human out of decomposition-based optimization via artificial intelligence, Part I: Learning when to decompose. Computers & Chemical Engineering, 186 (2024), p.108688.

[10]. Mitrai, I. and Daoutidis, P., Taking the human out of decomposition-based optimization via artificial intelligence, Part II: Learning to initialize. Computers & Chemical Engineering, 186 (2024), p.108686.

[11]. Vaupel, Y., Hamacher, N.C., Caspari, A., Mhamdi, A., Kevrekidis, I.G. and Mitsos, A., Accelerating nonlinear model predictive control through machine learning. Journal of Process Control, 92 (2020), pp.261-270.