Comparison of neural networks and fuzzy relational systems in dynamic modelling

Rizwan Saleem, Bruce Postlethwaite

Research output: Chapter in Book/Report/Conference proceedingChapter

Abstract

Both neural networks and fuzzy relational models show great potential for modelling poorly understood and highly non-linear systems. Recent papers have suggested that both techniques could be used to form the model in model-based controller designs. Although the two techniques can be targeted at the same role, they are fundamentally different. Fuzzy relational models attempt to capture relationships between qualitative states and therefore represent the type of qualitative models used in everyday commonsense reasoning. Neural networks instead try to imitate the hardware involved in thinking and can generate their results from the complex interactions between the separate network elements. Although the two methods are quite different in conception, the authors are not aware of any work that has been done to compare their performance in dynamic modelling, and this paper is an attempt to remedy this. In order to compare the two techniques, the well known Box-Jenkins furnace data were used. The software used to generate the neural networks was 'Neural-Works Explorer', and in-house software was used to generate the fuzzy relational models. The predictive performance of both methods was compared on the dataset for a variety of model configurations. The paper describes the results of these tests and discusses the effects of changing model parameters on predictive and practical performance. For the neural models the factors investigated were: network configuration, transfer function type, and various combinations of learning schemes. The factors considered for the fuzzy relational models were the model structure and reference set definitions. As well as predictive performance criteria, practical modelling criteria such as modelling time, sensitivity, etc, were also used to compare the two methods.
LanguageEnglish
Title of host publicationProceeding from the International Conference on Control 1994
PublisherIEEE
Pages1448-1452
Number of pages4
Volume1 & 2
ISBN (Print)0-85296-612-1
Publication statusPublished - Mar 1994

Fingerprint

Neural networks
Model structures
Transfer functions
Nonlinear systems
Furnaces
Hardware
Controllers

Keywords

  • neural networks
  • fuzzy relational systems
  • dynamic modelling

Cite this

Saleem, R., & Postlethwaite, B. (1994). Comparison of neural networks and fuzzy relational systems in dynamic modelling. In Proceeding from the International Conference on Control 1994 (Vol. 1 & 2, pp. 1448-1452). IEEE.
Saleem, Rizwan ; Postlethwaite, Bruce. / Comparison of neural networks and fuzzy relational systems in dynamic modelling. Proceeding from the International Conference on Control 1994. Vol. 1 & 2 IEEE, 1994. pp. 1448-1452
@inbook{6884bb92ae7e4538a4731e142ac11c0c,
title = "Comparison of neural networks and fuzzy relational systems in dynamic modelling",
abstract = "Both neural networks and fuzzy relational models show great potential for modelling poorly understood and highly non-linear systems. Recent papers have suggested that both techniques could be used to form the model in model-based controller designs. Although the two techniques can be targeted at the same role, they are fundamentally different. Fuzzy relational models attempt to capture relationships between qualitative states and therefore represent the type of qualitative models used in everyday commonsense reasoning. Neural networks instead try to imitate the hardware involved in thinking and can generate their results from the complex interactions between the separate network elements. Although the two methods are quite different in conception, the authors are not aware of any work that has been done to compare their performance in dynamic modelling, and this paper is an attempt to remedy this. In order to compare the two techniques, the well known Box-Jenkins furnace data were used. The software used to generate the neural networks was 'Neural-Works Explorer', and in-house software was used to generate the fuzzy relational models. The predictive performance of both methods was compared on the dataset for a variety of model configurations. The paper describes the results of these tests and discusses the effects of changing model parameters on predictive and practical performance. For the neural models the factors investigated were: network configuration, transfer function type, and various combinations of learning schemes. The factors considered for the fuzzy relational models were the model structure and reference set definitions. As well as predictive performance criteria, practical modelling criteria such as modelling time, sensitivity, etc, were also used to compare the two methods.",
keywords = "neural networks, fuzzy relational systems, dynamic modelling",
author = "Rizwan Saleem and Bruce Postlethwaite",
year = "1994",
month = "3",
language = "English",
isbn = "0-85296-612-1",
volume = "1 & 2",
pages = "1448--1452",
booktitle = "Proceeding from the International Conference on Control 1994",
publisher = "IEEE",

}

Saleem, R & Postlethwaite, B 1994, Comparison of neural networks and fuzzy relational systems in dynamic modelling. in Proceeding from the International Conference on Control 1994. vol. 1 & 2, IEEE, pp. 1448-1452.

Comparison of neural networks and fuzzy relational systems in dynamic modelling. / Saleem, Rizwan; Postlethwaite, Bruce.

Proceeding from the International Conference on Control 1994. Vol. 1 & 2 IEEE, 1994. p. 1448-1452.

Research output: Chapter in Book/Report/Conference proceedingChapter

TY - CHAP

T1 - Comparison of neural networks and fuzzy relational systems in dynamic modelling

AU - Saleem, Rizwan

AU - Postlethwaite, Bruce

PY - 1994/3

Y1 - 1994/3

N2 - Both neural networks and fuzzy relational models show great potential for modelling poorly understood and highly non-linear systems. Recent papers have suggested that both techniques could be used to form the model in model-based controller designs. Although the two techniques can be targeted at the same role, they are fundamentally different. Fuzzy relational models attempt to capture relationships between qualitative states and therefore represent the type of qualitative models used in everyday commonsense reasoning. Neural networks instead try to imitate the hardware involved in thinking and can generate their results from the complex interactions between the separate network elements. Although the two methods are quite different in conception, the authors are not aware of any work that has been done to compare their performance in dynamic modelling, and this paper is an attempt to remedy this. In order to compare the two techniques, the well known Box-Jenkins furnace data were used. The software used to generate the neural networks was 'Neural-Works Explorer', and in-house software was used to generate the fuzzy relational models. The predictive performance of both methods was compared on the dataset for a variety of model configurations. The paper describes the results of these tests and discusses the effects of changing model parameters on predictive and practical performance. For the neural models the factors investigated were: network configuration, transfer function type, and various combinations of learning schemes. The factors considered for the fuzzy relational models were the model structure and reference set definitions. As well as predictive performance criteria, practical modelling criteria such as modelling time, sensitivity, etc, were also used to compare the two methods.

AB - Both neural networks and fuzzy relational models show great potential for modelling poorly understood and highly non-linear systems. Recent papers have suggested that both techniques could be used to form the model in model-based controller designs. Although the two techniques can be targeted at the same role, they are fundamentally different. Fuzzy relational models attempt to capture relationships between qualitative states and therefore represent the type of qualitative models used in everyday commonsense reasoning. Neural networks instead try to imitate the hardware involved in thinking and can generate their results from the complex interactions between the separate network elements. Although the two methods are quite different in conception, the authors are not aware of any work that has been done to compare their performance in dynamic modelling, and this paper is an attempt to remedy this. In order to compare the two techniques, the well known Box-Jenkins furnace data were used. The software used to generate the neural networks was 'Neural-Works Explorer', and in-house software was used to generate the fuzzy relational models. The predictive performance of both methods was compared on the dataset for a variety of model configurations. The paper describes the results of these tests and discusses the effects of changing model parameters on predictive and practical performance. For the neural models the factors investigated were: network configuration, transfer function type, and various combinations of learning schemes. The factors considered for the fuzzy relational models were the model structure and reference set definitions. As well as predictive performance criteria, practical modelling criteria such as modelling time, sensitivity, etc, were also used to compare the two methods.

KW - neural networks

KW - fuzzy relational systems

KW - dynamic modelling

UR - http://ieeexplore.ieee.org/xpl/mostRecentIssue.jsp?punumber=1169

M3 - Chapter

SN - 0-85296-612-1

VL - 1 & 2

SP - 1448

EP - 1452

BT - Proceeding from the International Conference on Control 1994

PB - IEEE

ER -

Saleem R, Postlethwaite B. Comparison of neural networks and fuzzy relational systems in dynamic modelling. In Proceeding from the International Conference on Control 1994. Vol. 1 & 2. IEEE. 1994. p. 1448-1452