School of Civil and Environmental Engineering

Ph.D. Thesis Defense Announcement

Layered ARTIFICIAL ITELLIGENCE Predictive-Control Framework for Next-Generation Wastewater Treatment Plants

By Zhaoyang Xiong

Advisor:

Dr. Yongsheng Chen

Committee Members:  Dr. Xing Xie (CEE), DR. Joe Bozeman (CEE), DR. Ameet Pinto (CEE), DR. Enlu Zhou (ISYE)

Date and Time: December, 3, 2025. 3pm EST

Location: Daniel Lab 303

Virtual Link: Join the meeting now

Meeting ID: 281 877 299 879 6

Passcode: jX6DT3Aq

ABSTRACT
Municipal wastewater treatment plants (WWTPs) operate under changing conditions that do not always match the pace of their control systems. Flow and water quality can shift quickly, while biological processes and mechanical equipment respond more gradually. Most WWTPs rely on fixed setpoints and operator experience to bridge these gaps. This approach keeps effluent quality reliable, but it also limits WWTP’s ability to anticipate disturbances and adjust aeration and recirculation efficiently.
This dissertation introduces a three-layered artificial intelligence (AI) predictive control framework that works with the Supervisory Control and Data Acquisition (SCADA) and programmable logic controller (PLC) systems already in place at most WWTPs. The framework is organized into a Prediction Layer, a Local Control Layer, and a Plant-Wide Optimization Layer. Each layer is demonstrated through a full-scale field study at a municipal WWTP.
The first layer uses long-term operating records to forecast short-term effluent behavior. These forecasts give operators earlier awareness of shifts in loading conditions and help them adjust operation before large changes occur. The second layer focuses on dissolved oxygen (DO) control. A short-term DO predictor is paired with a rule-based decision module, allowing the controller to act in a steady and interpretable way. This design aligns more closely with how operators make decisions and avoids the abrupt actions that sometimes arise from fully data-driven controllers. The third layer coordinates multiple biological zones by combining predictive modeling with supervisory control embedded in SCADA. It provides a structured way to balance aeration, internal recirculation, external recirculation, and carbon dosing under changing influent conditions.
Together, these layers form a practical path toward more anticipatory and energy-conscious operation in municipal wastewater treatment. The framework improves decision-making without requiring major changes to existing infrastructure and supports a gradual transition toward smarter WWTP control.