2-D Defect Profile Reconstruction from Ultrasonic Guided Wave Signals Based on Radial Wavelet Basis Function Neural Network with ELM

Abstract

The reconstruction of defect profiles based on ultrasonic guided waves means the acquisition of defect profiles and parameters from ultrasonic guided wave signals. To achieve multi-resolution approximation, this paper proposed a reconstruction approach based on Radial Wavelet Basis Function Neural Network (RWBFNN), which combines wavelet analysis and neural network. Gaussian radial basis functions and Mexican hat wavelet frames are used as scaling functions and wavelet functions respectively. The training and testing samples contain simulation data and experimental data. The input data sets are defect echo signals, and the output data sets are 2-D profile parameters. To reduce the training time and simplify the profile reconstruction procedure without losing accuracy, Extreme Learning Machine (ELM) is adopted simultaneously. The results indicate that significant advantages can be obtained over other defect profile reconstruction schemes, and the accuracy of the predicted defect profile can be controlled by the resolution of the network with the lower computational complexity.