Advanced system for automating eddy-current nondestructive evaluation

Various nondestructive evaluation (NDE) techniques are widely used in the inspection of sub-surface flaws. This paper introduces the results of a research conducted to enhance the performance of eddy-current nondestructive evaluation (ECNDE) by developing an integrated computer based system. Advantages of this system include increasing test speed, while avoiding errors due to human factors. The system can be used to optimize various parameters affecting the performance of inspecting sub-surface cracks including probe configuration and operating frequency range. Inspectability is enhanced by advanced processing of raw eddy current signal. The two-dimensional wavelet transform is applied to eddy current c-scan images to extract feature vectors representing the material flaws. Artificial neural network techniques are then invoked to automate the detection and classification of sub-surface flaws.