A thesis to be submitted to the Department of Chemical Engineering for Masters in Chemical Engineering
Thesis time: Jan 2022 - May 2023
Thesis submission: Anticipated Dec 2024
With the growing automation and interconnectivity of industrial processes, the number of instrumentation components like actuators, sensors, and communication links is expanding. Consequently, the probability of faults or malfunctions in these components is increasing that leads to serious degradation in the closed-loop control quality and, if not properly handled, can even lead to a complete breakdown of the process operation. Therefore, developing systematic methods for handling faults is crucial to ensure safety, reliability, and profitability in networked process systems. This thesis presents a unified framework for real-time fault detection, estimation, and accommodation in large-scale networked process systems with multiple simultaneous sensor faults. By combining machine learning techniques with model-based control strategies, the proposed framework addresses the challenges of fault detection, estimation, and accommodation in single-unit and interconnected process networks.