Note: This content is accessible to all versions of every browser. However, this browser does not seem to support current Web standards, preventing the display of our site's design details.

  

Fast Plant Testing for MPC

Back
Abstract:
The widely accepted empirical way to characterize the dynamics of a system is to apply perturbation signals at the input channels and measure the response of the system to these signals. The input and output signals are then processed to give the required estimate of the dynamics of the system. This procedure is well known as system identification. In the vast literature on system identification relatively little attention has been paid to the design of the perturbation signals, other than to the fact that the signals should be persistently exciting. In this talk a method of plant testing and system identification with emphasis on test time reduction is presented. The plant testing time reduction is achieved via selection of binary multi-frequency (BMF) signals as input perturbation signals designed to excite the system in a control-relevant frequency range. The plant testing is conducted by means of a multivariable test with each input being excited with a delayed copy of the “mother” BMF signal. On average, the minimum testing time with multivariable BMF testing is 85-90% shorter than with conventional step testing. Once the test has been designed, its plant friendliness can be evaluated ahead of time via simulation of predictive models based on a priori knowledge of the system. The amplitudes of the perturbation signals used to excite the different input channels are optimized to maximize the output variability and the signal-to-noise ratio within pre-specified input and output constraints. The resulting data from plant testing is analyzed using parametric modeling techniques. * Talk based on the paper “ Novel Plant Test for MPC”, by Simone L. Kothare and Jorge A. Mandler, presented at Dycops 7 2004.

Type of Seminar:
Public Seminar
Speaker:
Dr. Jorge A. Mandler
GEO Advanced Control / Air Products and Chemicals, Inc. / Allentown, PA, USA
Date/Time:
Apr 15, 2005   17:00
Location:

ETH Zentrum, Physikstrasse 3, Building ETL, Room K 25
Contact Person:

Prof. M. Morari
File Download:

Request a copy of this publication.
Biographical Sketch:
Since joining Air Products in 1986, Dr. Mandler has worked in the broad areas of Advanced Control, from mathematical modeling and dynamic simulation, to system identification, to new technology evaluation, to the implementation of model predictive control and online optimization. Dr. Mandler currently leads a global effort on Advanced Control Research, working to find applications in non traditional areas, while at the same time continuing to provide solutions for the more traditional process areas. His work has been presented in national and international conferences, and has been awarded US and international patents. Throughout his career Dr. Mandler has maintained an active involvement with the academic control community by means of publications and by organizing modeling and control sessions for AIChE (CAST - Computing and Systems Technology Division) and the American Control Conference. He participated in CAST Division Programming from 1987 to 2001, chairing or co-chairing over 15 sessions. He served in the Program Committees of several international conferences. As a Director of CAST for the term 1997-1999 Dr. Mandler was involved in efforts aimed at increasing industrial participation in CAST activities. Dr. Mandler holds B.Sc. and M.Sc. degrees from the Technion, Israel Institute of Technology, and a Ph.D. from the California Institute of Technology, all in Chemical Engineering.