Executive Summary : | Process Systems Engineering (PSE) focuses on improving decision-making processes in the chemical supply chain, addressing the discovery, design, manufacture, and distribution of chemical products. Robust optimization strategies, such as worst-case analysis, are used to protect decision-makers from parameter uncertainty and stochastic uncertainty. Overconservatism is crucial in robust optimization when selecting the uncertain parameter set within which the worst case is estimated. Academics are aiming to recognize PSE research as a dynamic field with practical applications. Developing data-driven PSE techniques is one of the primary goals to bridge the gap between academia and business. This can be achieved using AI algorithms and systems engineering. The fundamental element of Robust Optimization (RO) is various uncertainty sets, such as interval and finite scenarios with distinct tractability and conservativeness. Currently, available techniques assume uncertainties in each dimension are independently and symmetrically distributed. The aim is to develop an effective data-driven approach for RO that is tailored to uncertainty set constructions and computational implementations, bridging machine learning and robust optimization directly. The plan involves implementing clustering and classification algorithms to create a data-driven uncertainty set, such as support vector machines, KNN, Logistic Regression, and Naive-Bayes. This research will allow users to control conservatism and complexity of RO problems, adapting the complexity of data intelligently. |