Development of neural network driven fuzzy controller for outlet sodium temperature of DHX
- Power Reactor and Nuclear Fuel Development Corp., Tsuruga, Fukui (Japan). Monju Construction Office
- Power Reactor and Nuclear Fuel Development Corp., Oarai, Ibaraki (Japan). Oarai Engineering Center
Fuzzy controls are capable to exquisitely control non-linear dynamic systems in wide operating range, using linguistic description to define the control law. However the selection and the definition of the fuzzy rules and sets require a tedious trial and error process based on experience. As a method to overcome this limitation, a neural network driven fuzzy control (NDF), where the learning capability of the neural network (NN) is used to build the fuzzy rules and sets, is presented in this paper. In the NDF control the IF part of a fuzzy control is represented by a multilayer NN while the THEN part is represented by a series of multilayer NNs which calculate the desirable control action. In this work the usual stepwise variable reduction method, used for the selection of the input variable in the THEN part NN, is replaced with a learning algorithm with forgetting mechanism that realizes the automatic reduction of the variables and the tuning up of all the fuzzy control law i.e. the membership function. The NDF has been successfully applied to control the outlet sodium temperature of a dump heat exchanger (DHX) of a FBR plant.
- OSTI ID:
- 271895
- Report Number(s):
- CONF-960306-; ISBN 0-7918-1226-X; TRN: 96:017204
- Resource Relation:
- Conference: ICONE 4: ASME/JSME international conference on nuclear engineering, New Orleans, LA (United States), 10-13 Mar 1996; Other Information: PBD: 1996; Related Information: Is Part Of ICONE-4: Proceedings. Volume 1 -- Part B: Basic technological advances; Rao, A.S. [ed.] [General Electric Nuclear Energy, San Jose, CA (United States)]; Duffey, R.B. [ed.] [Brookhaven National Lab., Upton, NY (United States)]; Elias, D. [ed.] [Commonwealth Edison, Downers Grove, IL (United States)]; PB: 564 p.
- Country of Publication:
- United States
- Language:
- English
Similar Records
Neural network-based self-organizing fuzzy controller for transient stability of multimachine power systems
Monitoring nuclear reactor systems using neural networks and fuzzy logic