Fault Detection and Diagnosis in Process Systems

With the widespread availability of Distributed Control Systems (DCS), continuous monitoring of chemical process operations is greatly facilitated. Plant operators are asked to manage the operation in such a way as to ensure optimal production levels, while attending occasional alarm situations that may result from equipment malfunctions. Timely identification of such abnormal situations may prove to be critical when there is a potential for a safety hazard that may affect not only the plant and its personnel but also the surrounding communities. It is imperative that human expertise is complemented by computerized support systems which consist of various data analysis and interpretation strategies that can provide guidance to the plant personnel for handling abnormal situations. Our group uses a number of techniques ranging from hidden Markov models and Principal Components Analysis to clustering to detect and diagnose faulty operations.

Our collaborator in this project is Professor J.A. Romagnoli (LSU).

Related Publications

• Beaver, S, J.A. Romagnoli and A. Palazoglu, "Cluster Analysis for Autocorrelated and Cyclic Chemical Process Data," Ind. & Eng. Chem. Research, 46, 3610-3622 (2007).

• Sun, W., A. Palazoglu, and J.A. Romagnoli, “Detection of Abnormal Process Trends Using Wavelet-Domain Hidden-Markov Models,” AIChE J., 49, 140-150 (2003).