Application of neural networks to laser ultrasonic NDE of bonded structures

The purpose of this study is to develop artificial neural networks for the nondestructive evaluation of a bounded structure using laser generated surface waves. The radial basis model is employed and the inputs of the networks are the sampled dispersion curves while the output is the thickness of the bonding layer or unbond ratio. Numerical simulated dispersion curves with noises were used to train and test the neural networks. The dispersion curves of laser ultrasonic experiments acquired by spectral analysis method were also utilized to test the systems. For comparison, simplex inverse method was also introduced. The results demonstrate that neural networks are well suited for on-line evaluation of bonding thickness and unbond ratio of a bonded layered structure.