MACHINE LEARNING AIDED MULTI-OBJECTIVE OPTIMIZATION AND MULTI-CRITERIA DECISION MAKING IN CHEMICAL ENGINEERING

Abstract

To bridge the existing research gaps in the field of chemical engineering, this thesis makes the following contributions. Firstly, it introduces the Preference Ranking On the Basis of Ideal-average Distance (PROBID) and simpler PROBID (sPROBID) methods for multi-criteria decision making (MCDM), which are more consistent and robust, and can seamlessly connect with multi-objective optimization (MOO) in chemical engineering applications. Secondly, it proposes fuzzy PROBID and fuzzy sRPOBID methods to handle broader decision-making applications in chemical engineering under a fuzzy environment. Thirdly, it proposes a comprehensive machine learning (ML) aided MOO-MCDM framework to accelerate data-driven studies for optimization applications in chemical engineering. Lastly, by integrating the proposed ML aided MOO-MCDM framework with model predictive control (MPC), this thesis enhances the functionality of MPC to efficiently handle multiple objectives and leverage ML models in process control.