Feedback linearizing control using hybrid neural networks identified by sensitivity approach

Janos Madar, J. Abonyi, F. Szeifert

Research output: Contribution to journalArticle

20 Citations (Scopus)

Abstract

Globally linearizing control (GLC) is a control algorithm capable of using nonlinear process model directly. In GLC, mostly, first-principle models derived from dynamic mass, energy and momentum balances are used. When the process is not perfectly known, the unknown parts of the first principles model should be represented by black-box models, e.g. by neural networks. This paper is devoted to the identification and application of such hybrid models for GLC. It is shown that the first principles part of the model determines the dominant structure of the controller, while the black-box elements of the hybrid model are used as state and/or disturbance estimators. For the identification of the neural network elements of the hybrid model a sensitivity approach based algorithm has been developed. The underlying framework is illustrated by the temperature control of a continuous stirred tank reactor where a neural network is used to model the heat released by an exothermic chemical reaction.

Original languageEnglish
Pages (from-to)343-351
Number of pages9
JournalEngineering Applications of Artificial Intelligence
Volume18
Issue number3
DOIs
Publication statusPublished - Apr 2005

Fingerprint

Feedback control
Neural networks
Temperature control
Chemical reactions
Identification (control systems)
Momentum
Controllers

Keywords

  • Globally linearizing control
  • Neural network
  • Semi-mechanistic models
  • Sensitivity approach

ASJC Scopus subject areas

  • Artificial Intelligence
  • Control and Systems Engineering

Cite this

@article{862522d06e1c49649bcdd2667895bc26,
title = "Feedback linearizing control using hybrid neural networks identified by sensitivity approach",
abstract = "Globally linearizing control (GLC) is a control algorithm capable of using nonlinear process model directly. In GLC, mostly, first-principle models derived from dynamic mass, energy and momentum balances are used. When the process is not perfectly known, the unknown parts of the first principles model should be represented by black-box models, e.g. by neural networks. This paper is devoted to the identification and application of such hybrid models for GLC. It is shown that the first principles part of the model determines the dominant structure of the controller, while the black-box elements of the hybrid model are used as state and/or disturbance estimators. For the identification of the neural network elements of the hybrid model a sensitivity approach based algorithm has been developed. The underlying framework is illustrated by the temperature control of a continuous stirred tank reactor where a neural network is used to model the heat released by an exothermic chemical reaction.",
keywords = "Globally linearizing control, Neural network, Semi-mechanistic models, Sensitivity approach",
author = "Janos Madar and J. Abonyi and F. Szeifert",
year = "2005",
month = "4",
doi = "10.1016/j.engappai.2004.09.006",
language = "English",
volume = "18",
pages = "343--351",
journal = "Engineering Applications of Artificial Intelligence",
issn = "0952-1976",
publisher = "Elsevier Limited",
number = "3",

}

TY - JOUR

T1 - Feedback linearizing control using hybrid neural networks identified by sensitivity approach

AU - Madar, Janos

AU - Abonyi, J.

AU - Szeifert, F.

PY - 2005/4

Y1 - 2005/4

N2 - Globally linearizing control (GLC) is a control algorithm capable of using nonlinear process model directly. In GLC, mostly, first-principle models derived from dynamic mass, energy and momentum balances are used. When the process is not perfectly known, the unknown parts of the first principles model should be represented by black-box models, e.g. by neural networks. This paper is devoted to the identification and application of such hybrid models for GLC. It is shown that the first principles part of the model determines the dominant structure of the controller, while the black-box elements of the hybrid model are used as state and/or disturbance estimators. For the identification of the neural network elements of the hybrid model a sensitivity approach based algorithm has been developed. The underlying framework is illustrated by the temperature control of a continuous stirred tank reactor where a neural network is used to model the heat released by an exothermic chemical reaction.

AB - Globally linearizing control (GLC) is a control algorithm capable of using nonlinear process model directly. In GLC, mostly, first-principle models derived from dynamic mass, energy and momentum balances are used. When the process is not perfectly known, the unknown parts of the first principles model should be represented by black-box models, e.g. by neural networks. This paper is devoted to the identification and application of such hybrid models for GLC. It is shown that the first principles part of the model determines the dominant structure of the controller, while the black-box elements of the hybrid model are used as state and/or disturbance estimators. For the identification of the neural network elements of the hybrid model a sensitivity approach based algorithm has been developed. The underlying framework is illustrated by the temperature control of a continuous stirred tank reactor where a neural network is used to model the heat released by an exothermic chemical reaction.

KW - Globally linearizing control

KW - Neural network

KW - Semi-mechanistic models

KW - Sensitivity approach

UR - http://www.scopus.com/inward/record.url?scp=14844282841&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=14844282841&partnerID=8YFLogxK

U2 - 10.1016/j.engappai.2004.09.006

DO - 10.1016/j.engappai.2004.09.006

M3 - Article

AN - SCOPUS:14844282841

VL - 18

SP - 343

EP - 351

JO - Engineering Applications of Artificial Intelligence

JF - Engineering Applications of Artificial Intelligence

SN - 0952-1976

IS - 3

ER -