W. Yan et Br. Upadhyaya, AN INTEGRATED SIGNAL-PROCESSING AND NEURAL NETWORKS SYSTEM FOR STEAM-GENERATOR TUBING DIAGNOSTICS USING EDDY-CURRENT INSPECTION, Annals of nuclear energy, 23(10), 1996, pp. 813-825
The primary purpose of this research was to develop an integrated appr
oach by combining information compression methods and artificial neura
l networks for the monitoring of plant components using nondestructive
evaluation (NDE) data. Specifically, data from eddy current inspectio
n of steam generator tubing were utilized to evaluate this technology.
The focus of the research was to develop and test various data compre
ssion methods (for eddy current data) and the performance of different
neural network paradigms for defect classification and defect paramet
er estimation. Feedforward, fully-connected neural networks, that use
the back-propagation algorithm for network training, were implemented
for defect classification and defect parameter estimation using a modu
lar network architecture. A large eddy current tube inspection databas
e was acquired from the Metals and Ceramics Division of Oak Ridge Nati
onal Laboratory (ORNL). These data were used to study the performance
of artificial neural networks for defect type classification and for e
stimating defect parameters. Most of the study was made using the Neur
alWorks Professional II/Plus software. A PC-based data pre-processing
and display program was also developed as part of an expert system for
data management and decision making. The results of the analysis show
ed that for effective (low-error) defect classification and estimation
of parameters, it is necessary to identify proper feature vectors usi
ng different data representation methods. The integration of data comp
ression and artificial neural networks for information processing was
established as an efficient technique for automation of diagnostics us
ing nondestructive evaluation methods.