Closed-loop identification and predictive control of chemical processes /

Model Predictive Control (MPC) is currently the most

Bibliographic Details
Main Author: Shouche, Manoj S., 1970-
Format: Thesis Book
Language:English
Published: [Place of publication not identified] : [publisher not identified] ; 1996.
Subjects:
Online Access:http://proxy.library.tamu.edu/login?url=http://proquest.umi.com/pqdweb?did=739667721&sid=1&Fmt=2&clientId=2945&RQT=309&VName=PQD
Description
Summary:Model Predictive Control (MPC) is currently the most
effective control scheme in the process industry. The
popularity of MPC lies in the fact that it can handle process
input/output constraints systematically during the design and
the implementation of the controller. MPC relies on a
process model to generate optimal inputs by performing an on-
line optimization. The process model is usually made
available at the controller design stage by performing open-
loop experiments on the process. The closed-loop performance
depends on the accuracy of the model and may become
unacceptable as the modeling error increases. Hence, due to
safety and performance reasons, it may become necessary to
update the process model under closed-loop conditions. In
this work, a new approach to simultaneous constrained Model
Predictive Control and Identification (WPCI) is developed.
In this approach, attempts are made to update transfer
function process models with minimum deterioration in the
control performance while satisfying the process input/output
constraints. The method relies on the persistence of
excitation condition, which is used to produce dynamic
information about the process under feedback control. The
persistence of excitation
condition, a function of the process inputs and the
process outputs, is first converted into a condition
involving only the process inputs. The resulting new
condition on the inputs is used as an additional
constraint in the conventional MPC formulation. The
resulting non-convex on-line optimization problem is
transformed into a convex problem; the solution of which
is obtained by solving a series of converging semidefinite
programming problems. A deterministic global optimization
technique, based on the branch-and-bound approach is also
developed to obtain the global solution of the on-line
optimization problem The global solution can be used to
measure the performance of the semidefinite programming
approach. The applicability of MPCI to update dynamic
process models is demonstrated through various
simulations. The results obtained show that MPCI gives
superior performance as compared to the traditional
methods in terms of (a) handling process input/output
constraints, and (b) keeping the deterioration of the
output regulation to a minimum value, while performing
closed-loop identification.
Item Description:Vita.
"Major Subject: Chemical Engineering".
Physical Description:xiii, 162 leaves : illustrations ; 28 cm.
Issued also on microfiche from University Microfilms Inc.
Bibliography:Includes bibliographical references: pages 133-138.