You need JavaScript to view this

Dynamic optimization in state-space predictive control schemes

Abstract

Optimization is a useful and in many cases necessary tool to increase the profit rates from an industrial process. This will often lead to process operation at the intersection of constraints. A multivariable controller which has a flexible structure and explicitly accounts for the constraints is needed for the realization of such operation points. Only the model-predictive controllers have succeeded in this task, and such schemes have been developed and implemented during the last two decades. In the presentation of proven model predictive control (MPC), most emphasis is given to the Dynamic Matrix Control (DMC) developed by the Shell Oil Company during the seventies. All the known implementations of MPC make use of a linear input-output model and quadratic control objectives. The appealing consequences are that the resulting quadratic program can be solved at an acceptable speed by modest computing power, and that the model can be established experimentally. Constraints on manipulated and measured variables are explicitly accounted for. The State-Space Predictive Control (SSPC) scheme which is proposed in this thesis offers certain advantages to the earlier MPC-techniques, at the cost if increased modeling efforts and computational loads. The generally nonlinear statespace model is favorable to a linear description when  More>>
Authors:
Publication Date:
Dec 31, 1991
Product Type:
Thesis/Dissertation
Report Number:
NEI-NO-201
Reference Number:
SCA: 320300; PA: NW-92:005012; SN: 92000720030
Resource Relation:
Other Information: TH: Thesis (Dr.ing.).; PBD: 1991
Subject:
32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATION; INDUSTRIAL PLANTS; OPTIMIZATION; ENERGY CONSERVATION; ECONOMICS; MATHEMATICAL MODELS; CONTROL THEORY; COMPUTERIZED CONTROL SYSTEMS; 320300; INDUSTRIAL AND AGRICULTURAL PROCESSES
OSTI ID:
10141615
Research Organizations:
Trondheim Univ. (Norway). Norges Tekniske Hoegskole
Country of Origin:
Norway
Language:
English
Other Identifying Numbers:
Other: ON: DE92506527; ISBN 82-7119-265-5; TRN: NO9205012
Availability:
OSTI; NTIS (US Sales Only)
Submitting Site:
NW
Size:
289 p.
Announcement Date:
Jul 05, 2005

Citation Formats

Strand, S. Dynamic optimization in state-space predictive control schemes. Norway: N. p., 1991. Web.
Strand, S. Dynamic optimization in state-space predictive control schemes. Norway.
Strand, S. 1991. "Dynamic optimization in state-space predictive control schemes." Norway.
@misc{etde_10141615,
title = {Dynamic optimization in state-space predictive control schemes}
author = {Strand, S}
abstractNote = {Optimization is a useful and in many cases necessary tool to increase the profit rates from an industrial process. This will often lead to process operation at the intersection of constraints. A multivariable controller which has a flexible structure and explicitly accounts for the constraints is needed for the realization of such operation points. Only the model-predictive controllers have succeeded in this task, and such schemes have been developed and implemented during the last two decades. In the presentation of proven model predictive control (MPC), most emphasis is given to the Dynamic Matrix Control (DMC) developed by the Shell Oil Company during the seventies. All the known implementations of MPC make use of a linear input-output model and quadratic control objectives. The appealing consequences are that the resulting quadratic program can be solved at an acceptable speed by modest computing power, and that the model can be established experimentally. Constraints on manipulated and measured variables are explicitly accounted for. The State-Space Predictive Control (SSPC) scheme which is proposed in this thesis offers certain advantages to the earlier MPC-techniques, at the cost if increased modeling efforts and computational loads. The generally nonlinear statespace model is favorable to a linear description when the nonlinearities in the process dynamics are severe in the actual operating region. This mean that the advantage of the nonlinear model will be pronounced when the operating points are frequently changing. 32 figs., 18 tabs., 92 refs.}
place = {Norway}
year = {1991}
month = {Dec}
}