抄録
This paper describes non-destructive evaluation by means of the ultrasonic inspection of the frictional welding in cast iron using a neural network. Specimens altering the frictional welding condition were carried out to the inspection. Feature extracted waveforms in terms of the reflected echo from frictional welding location were learned and classified by the neural network. It was concluded that discriminations of frictional welding condition were possible by the proposed approach.
