Embedded Files
The operation of chemical processes is affected by multiple disturbances, such as changes in product demand and in the inlet conditions of the process. Rolling horizon-based strategies, such as Model Predictive Control, have been widely used to handle disturbances and guarantee safe and optimal operation. The efficiency of such strategies depends on the efficient repeated solution of optimization problems. Our group aims to develop computationally efficient solution methods by 1) exploiting the underlying structure of such problems, 2) developing tractable reduced-order models, and 3) combining optimization theory with artificial intelligence.
Some snippets of our previous work are presented below.
Machine Leaning-Based Initialization of Generalized Benders Decomposition for Mixed Integer Model Predictive Control
Machine Leaning-Based Initialization of Generalized Benders Decomposition for Mixed Integer Model Predictive Control
Selected Publications
Mitrai, I., and Daoutidis, P., 2023, Learning to recycle Benders cuts for Mixed Integer Model Predictive Control, under review.
Mitrai, I. and Daoutidis, P., 2024. Computationally efficient solution of mixed integer model predictive control problems via machine learning aided Benders Decomposition. Journal of Process Control, 137, p.103207. [arXiv][paper]
Mitrai, I., and Daoutidis, P., 2023, Machine Leaning-Based Initialization of Generalized Benders Decomposition for Mixed Integer Model Predictive Control, under review.
Mitrai, I., and Daoutidis, P., 2022, A multicut Generalized Benders Decomposition approach for the integration of process operations and dynamic optimization for continuous systems, Computers and Chemical Engineering, p.107859 [paper]
Mitrai, I., and Daoutidis, P., 2021, Efficient solution of Enterprise-wide Optimization Problems using Nester Stochastic Blockmodeling, Industrial & Engineering Chemistry Research, 60(40), pp. 14476-14494. [paper]
Page updated
Google Sites
Report abuse