(92d) Experimental Test Design for Mimo Identification Subject to Integral Controllability
AIChE Spring Meeting and Global Congress on Process Safety
2008
2008 Spring Meeting & 4th Global Congress on Process Safety
11th Topical on Refinery Processing - Jointly Co-sponsored with ACS
Control and Optimization in Refining I
Tuesday, April 8, 2008 - 10:00am to 10:30am
Model-based multivariable control requires good-quality process models which, in addition to standard specifications for model quality, also satisfy the integral controllability condition. To identify such models it is necessary to design appropriate identification experiments. In recent work we developed a rigorous methodology for the design of identification experiments that produce models satisfying the integral controllability condition. This methodology encompasses well known experiment designs but also suggests many new designs of practical importance. In this presentation we focus on what is a general challenge for experiment design methods, namely the dependence of the design on the very model to be identified. We demonstrate that a simple adaptive approach can produce reasonable results. We illustrate the proposed approach through simulations on a fluid catalytic crackling unit (FCCU) example and compare it to known experiment design schemes.