Neural nets and eddy-current testing
Artificial neural networks of a novel type have been trained and tested on a variety of eddy-current flaw signals commonly occurring in nuclear reactor steam generators with the ultimate goal of emulating, at least crudely, the vision and reasoning capabilities of the human analyst. The network methodology itself was that of Allen and Schell developed originally for studies of such biologically relevant neural properties as cognitive complementarity and concept formation. Because they are so important to the results obtained, we discuss the general characteristics of the approach in the Preamble. In Section I we describe the relevant aspects of the neural network configuration presently in use and in Section II the method by which the artificial neural systems have been trained. Finally, we discuss results of the training process for systems explored.
- Research Organization:
- Oak Ridge National Lab., TN (USA)
- Sponsoring Organization:
- DOE/ER
- DOE Contract Number:
- AC05-84OR21400
- OSTI ID:
- 6686399
- Report Number(s):
- CONF-900791-1; ON: DE90016044
- Country of Publication:
- United States
- Language:
- English
Similar Records
Reconstruction of Flaw Profiles Using Neural Networks and Multi-Frequency Eddy Current System
A practical guide to neural nets
Related Subjects
220000 -- Nuclear Reactor Technology
99 GENERAL AND MISCELLANEOUS
990200* -- Mathematics & Computers
BOILERS
EDDY CURRENT TESTING
ELECTROMAGNETIC TESTING
MATERIALS TESTING
NEURAL NETWORKS
NONDESTRUCTIVE TESTING
NUCLEAR FACILITIES
NUCLEAR POWER PLANTS
POWER PLANTS
STEAM GENERATORS
TESTING
THERMAL POWER PLANTS
TRAINING
VAPOR GENERATORS