Implementation of a fuzzy logic/neural network multivariable controller
Abstract
This paper describes a multivariable controller developed at the Idaho National Engineering Laboratory (INEL) that incorporates both fuzzy logic rules and a neural network. The controller was implemented in a laboratory demonstration and was robust, producing smooth temperature and water level response curves with short time constants. In the future, intelligent control systems will be a necessity for optimal operation of autonomous reactor systems located on earth or in space. Even today, there is a need for control systems that adapt to the changing environment and process. Hybrid intelligent control systems promise to provide this adaptive capability. Fuzzy logic implements our imprecise, qualitative human reasoning. The values of system variables (controller inputs) and control variables (controller outputs) are described in linguistic terms and subdivided into fully overlapping value ranges. The fuzzy rule base describes how combinations of input parameter ranges determine the output control values. Neural networks implement our human learning. In this controller, neural networks were embedded in the software to explore their potential for adding adaptability.
 Authors:

 Idaho National Engineering Lab., Idaho Falls (United States)
 Publication Date:
 OSTI Identifier:
 6670943
 Report Number(s):
 CONF921102
Journal ID: ISSN 0003018X; CODEN: TANSAO
 Resource Type:
 Conference
 Journal Name:
 Transactions of the American Nuclear Society; (United States)
 Additional Journal Information:
 Journal Volume: 66; Conference: Joint American Nuclear Society (ANS)/European Nuclear Society (ENS) international meeting on fifty years of controlled nuclear chain reaction: past, present, and future, Chicago, IL (United States), 1520 Nov 1992; Journal ID: ISSN 0003018X
 Country of Publication:
 United States
 Language:
 English
 Subject:
 22 GENERAL STUDIES OF NUCLEAR REACTORS; ELECTRIC CONTROLLERS; EXPERT SYSTEMS; FUZZY LOGIC; NEURAL NETWORKS; REACTOR CONTROL SYSTEMS; OPTIMIZATION; COMPUTER CODES; CONTROL EQUIPMENT; CONTROL SYSTEMS; EQUIPMENT; MATHEMATICAL LOGIC; 220400*  Nuclear Reactor Technology Control Systems
Citation Formats
Cordes, G A, Clark, D E, Johnson, J A, Smartt, H B, Wickham, K L, and Larson, T K. Implementation of a fuzzy logic/neural network multivariable controller. United States: N. p., 1992.
Web.
Cordes, G A, Clark, D E, Johnson, J A, Smartt, H B, Wickham, K L, & Larson, T K. Implementation of a fuzzy logic/neural network multivariable controller. United States.
Cordes, G A, Clark, D E, Johnson, J A, Smartt, H B, Wickham, K L, and Larson, T K. Wed .
"Implementation of a fuzzy logic/neural network multivariable controller". United States.
@article{osti_6670943,
title = {Implementation of a fuzzy logic/neural network multivariable controller},
author = {Cordes, G A and Clark, D E and Johnson, J A and Smartt, H B and Wickham, K L and Larson, T K},
abstractNote = {This paper describes a multivariable controller developed at the Idaho National Engineering Laboratory (INEL) that incorporates both fuzzy logic rules and a neural network. The controller was implemented in a laboratory demonstration and was robust, producing smooth temperature and water level response curves with short time constants. In the future, intelligent control systems will be a necessity for optimal operation of autonomous reactor systems located on earth or in space. Even today, there is a need for control systems that adapt to the changing environment and process. Hybrid intelligent control systems promise to provide this adaptive capability. Fuzzy logic implements our imprecise, qualitative human reasoning. The values of system variables (controller inputs) and control variables (controller outputs) are described in linguistic terms and subdivided into fully overlapping value ranges. The fuzzy rule base describes how combinations of input parameter ranges determine the output control values. Neural networks implement our human learning. In this controller, neural networks were embedded in the software to explore their potential for adding adaptability.},
doi = {},
url = {https://www.osti.gov/biblio/6670943},
journal = {Transactions of the American Nuclear Society; (United States)},
issn = {0003018X},
number = ,
volume = 66,
place = {United States},
year = {1992},
month = {1}
}