Degree

Doctor of Philosophy (PhD)

Department

Chemical Engineering

Document Type

Dissertation

Abstract

This dissertation explores the application of machine learning in chemical and biological systems, focusing on data-driven process modeling, reaction optimization, and control using supervised learning, unsupervised learning, and deep reinforcement learning (DRL). It presents three case studies demonstrating the integration of advanced machine learning techniques into generative design, bioprocess optimization, and bioinformatics.

In the first study, DRL is employed to design surfactant molecules with low critical micelle concentration (CMC), optimizing molecular structures for targeted properties. The second study applies Physics-Informed Neural Networks (PINNs) for the biokinetic modeling and optimization of a fed-batch bioreactor used for bio-product accumulation in cyanobacteria Plectonema. The PINN model exhibits strong interpolation and extrapolation capabilities, while DRL-based control enhances c-Phycocyanin yield by 78%, demonstrating the efficiency of AI-driven bioprocess optimization. The third study utilizes deep learning and unsupervised learning to cluster large-scale single-cell RNA sequencing data, facilitating bioinformatics-driven insights into cellular heterogeneity.

This research underscores the transformative potential of machine learning in optimizing reaction systems and process control. RL-based optimization enables efficient decision-making in molecular design, bioreactor regulation, and chemical process modeling, offering a robust AI-driven framework for chemical and biological engineering. These findings highlight the impact of deep learning in predictive modeling, process efficiency, and intelligent decision-making within complex manufacturing and biological systems.

Date

4-1-2025

Committee Chair

Benton, Michael G.

Download

Available for download on Tuesday, March 30, 2032

Share

COinS