Einladung zum Vortrag im Kolloquium Technische Kybernetik

Plant-wide Oscillation Detection and Diagnosis
- A Multivariate-Univariate Approach

Dr. Arun K. Tangirala
Department of Chemical Engineering
Indian Institute of Technology Madras
Chennai
India

    Zeit: Dienstag 23. 09. 2008 · 11:30 Uhr
    Ort: IST-Seminarraum 3.241 · Pfaffenwaldring 9 · Campus Stuttgart-Vaihingen

Abstract

The plant-wide oscillation detection and diagnosis problem is of considerable value and importance in process industry where hundreds to thousands of control loops are in place. We break up the problem of interest into two parts: (i) first identify the loops which have similar oscillatory characteristics and (ii) diagnose the source(s) of oscillations in each loop. The first step is enabled by setting up the multivariate source separation problem and solving it using the technique of non-negative matrix factorization (NMF). Modifications of the NMF algorithm in detecting the physical source signature will be presented. The subsequent step of diagnosis of source(s) of oscillations is carried out using specific univariate techniques. Particular attention will be paid to the detection of non-linearities due to valve stiction using the technique of Hilbert Huang Transform. Case studies to demonstrate the utility of the proposed methodologies will be presented. 

Biographical Information

Arun Tangirala obtained his Bachelors of Technology in Chemical Engineering from IIT Madras, India in 1996 and subsequently his Ph.D. in Process Control and Monitoring from the Dept. of Chemical & Materials Engineering, University of Alberta, Edmonton, Canada. After a three-year stint as a Research Manager at the University of Alberta, Arun Tangirala joined the Department of Chemical Engineering, IIT Madras as an Assistant Professor where he is currently employed. His research areas include the areas of Systems Engineering and Control, Closed-loop performance assessment, Multivariate Data Analysis, Time-Frequency Analysis of Process Systems, Control and Modelling of Fuel Cell Systems. He is currently guiding three doctoral and three masters students and is involved in teaching courses on Control, Identification and Applied Time-Series Analysis.