Bayesian Methods in Control Engineering
Feb.25 Tuesday @CSC 304 14:00~15:00
Bayesian approach, due to its mathematical rigor and flexibility in applications, has attracted increasing interest from both academia and practitioner. Bayesian theory can involve complex mathematical expressions. Yet the end result provides very meaningful solutions to practical problems. Although the control community may not be familiar with the term “Bayesian inference”, it has been unconsciously adopted by control scientists as early as the start of modern control. The best known application of Bayesian theory in control engineering is Kalman filter which has been widely adopted by the control community. It is now commonly recognized that many control related problems can be formulated under Bayesian framework. This presentation will give a historical overview of Bayesian methods in control engineering, current activities, and future trends. These will include system identification, state estimation, advanced filtering, fault detection & isolation, control performance monitoring & diagnosis, and soft sensors development.
Real-time predictive inferential sensing in the presence of uncertainties
Feb.27 Thursday @CSC 304 14:00~15:00
Operation of modern process industries is both a costly and technically complex business. It is of practical interest to investigate novel techniques to improve profitability while diligently maintaining environmental compliance. One of the proven approaches for finding solutions to achieve this objective is to develop innovative strategies for advanced monitoring and control of plant operations. Development and implementation of advanced monitoring and control techniques require real-time inference of critical process variables. However, on-line acquisition of such variables may involve difficulties due to the inadequacy of measurement techniques or low reliability of measuring devices. To overcome the shortcomings of traditional instrumentation, predictive inference has been designed to infer critical variables from real-time measurable secondary process variables. Predictive inference has become an emerging technology that has shown great potential in filling in the technological and financial gaps with little or no capital cost required. However, each inference is unique and there is no universal solution to the predictive inference problems. Hence, the novelty is reflected essentially in the solution strategies developed as each application poses its own challenges. Development of predictive inference mainly consists of four steps: 1) modeling, 2) prediction, 3) implementation and 4) monitoring. The main challenges are uncertainties involved in the development of predictive inference including uncertainty in data quality, in model parameters, in reference data and in operating conditions. These challenges call for establishment of a rigorous mathematical framework and practical rules. In this presentation, a general introduction to the main steps involved in predictive inference, mathematical principles behind robust modeling and inference, approaches to dealing with uncertainties and practical implementation is provided. The main challenges are discussed and solution strategies are illustrated through a number of successful applications.
Biao Huang (PhD, PEng) is Fellow of the Canadian Academy of Engineering, Professor of Department of Chemical and Materials Engineering at the University of Alberta, NSERC Senior Industrial Research Chair in Control of Oil Sands Processes, Alberta Innovates Technology Futures Industry Chair in Process Control, and Fellow of the Chemical Institute of Canada. He obtained a PhD degree in Process Control from the University of Alberta, in 1997. He received MSc degree (1986) and BSc degree (1983) in Automatic Control from the Beijing University of Aeronautics and Astronautics. Biao Huang joined the University of Alberta in 1997 as an Assistant Professor, promoted to Associate Professor in 2001, and full Professor in 2003. He is a recipient of Germany’s Alexander von Humboldt Research Fellowship, Canadian Chemical Engineer Society’s Syncrude Canada Innovation Award, APEGA’s Summit Award in Research Excellence, and many other awards. He has published 4 books and over 200 peer-reviewed journal papers, and had over 140 presentations in international conferences. His monograph on control performance assessment has become a seminal work and has enjoyed applications in chemical, petrochemical, oil & gas, mineral processing, and pulp & paper industries worldwide. He is currently the Deputy Editor-in-Chief for Control Engineering Practice, an associate editor for Canadian Journal of Chemical Engineering, and associate editor for Journal of Process Control.