Sensor validation in power plants using adaptive backpropagation neural network
Abstract
Signal validation and process monitoring problems in many cases require the prediction of one or more process variables in a system. The feasibility of using neural networks to characterize one variable as a function of other related variables is studied. The Backpropagation Network (BPN) is used to develop models'' of signals from both a commercial power plant and the EBR-II. Several innovations are made in the algorithm, the most significant of which is the progressive adjustment of the sigmoidal threshold function and weight updating terms, thus leading to the designation Adaptive'' Backpropagation Neural Network. The estimation of system variables is performed traditionally using either physical models or empirical models. The prediction of system variables is important in control systems for validating instrumentation outputs and for process monitoring. The model-based prediction assumes a fixed structure for characterizing steady-state or dynamic relationship among process variables. The applications to large and complex systems require more time in order to get an accurate model. Since our goal is to relate signals in a subsystem of a plant, such a relationship can be developed by using neural network models'' which provide results faster than model-based techniques. Both steady-state and transient behavior can be incorporated intomore »
- Authors:
- Publication Date:
- Research Org.:
- Tennessee Univ., Knoxville, TN (United States). Dept. of Nuclear Engineering
- Sponsoring Org.:
- USDOE; USDOE, Washington, DC (United States)
- OSTI Identifier:
- 6978406
- Report Number(s):
- CONF-900143-39
ON: DE93002959
- DOE Contract Number:
- FG07-88ER12824
- Resource Type:
- Conference
- Resource Relation:
- Conference: Institute for Electronic and Electrical Engineers (IEEE) nuclear science symposium, San Francisco, CA (United States), 15-19 Jan 1990
- Country of Publication:
- United States
- Language:
- English
- Subject:
- 21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS; 22 GENERAL STUDIES OF NUCLEAR REACTORS; 99 GENERAL AND MISCELLANEOUS//MATHEMATICS, COMPUTING, AND INFORMATION SCIENCE; EBR-2 REACTOR; REACTOR MONITORING SYSTEMS; NEURAL NETWORKS; NUCLEAR POWER PLANTS; SIGNAL CONDITIONING; TRAINING; VALIDATION; BREEDER REACTORS; EDUCATION; EPITHERMAL REACTORS; EXPERIMENTAL REACTORS; FAST REACTORS; FBR TYPE REACTORS; LIQUID METAL COOLED REACTORS; LMFBR TYPE REACTORS; NUCLEAR FACILITIES; POWER PLANTS; POWER REACTORS; REACTORS; RESEARCH AND TEST REACTORS; SODIUM COOLED REACTORS; TESTING; THERMAL POWER PLANTS; 220600* - Nuclear Reactor Technology- Research, Test & Experimental Reactors; 220400 - Nuclear Reactor Technology- Control Systems; 990200 - Mathematics & Computers
Citation Formats
Eryurek, E, and Upadhyaya, B R. Sensor validation in power plants using adaptive backpropagation neural network. United States: N. p., 1990.
Web.
Eryurek, E, & Upadhyaya, B R. Sensor validation in power plants using adaptive backpropagation neural network. United States.
Eryurek, E, and Upadhyaya, B R. 1990.
"Sensor validation in power plants using adaptive backpropagation neural network". United States. https://www.osti.gov/servlets/purl/6978406.
@article{osti_6978406,
title = {Sensor validation in power plants using adaptive backpropagation neural network},
author = {Eryurek, E and Upadhyaya, B R},
abstractNote = {Signal validation and process monitoring problems in many cases require the prediction of one or more process variables in a system. The feasibility of using neural networks to characterize one variable as a function of other related variables is studied. The Backpropagation Network (BPN) is used to develop models'' of signals from both a commercial power plant and the EBR-II. Several innovations are made in the algorithm, the most significant of which is the progressive adjustment of the sigmoidal threshold function and weight updating terms, thus leading to the designation Adaptive'' Backpropagation Neural Network. The estimation of system variables is performed traditionally using either physical models or empirical models. The prediction of system variables is important in control systems for validating instrumentation outputs and for process monitoring. The model-based prediction assumes a fixed structure for characterizing steady-state or dynamic relationship among process variables. The applications to large and complex systems require more time in order to get an accurate model. Since our goal is to relate signals in a subsystem of a plant, such a relationship can be developed by using neural network models'' which provide results faster than model-based techniques. Both steady-state and transient behavior can be incorporated into the network during training.},
doi = {},
url = {https://www.osti.gov/biblio/6978406},
journal = {},
number = ,
volume = ,
place = {United States},
year = {Mon Jan 01 00:00:00 EST 1990},
month = {Mon Jan 01 00:00:00 EST 1990}
}