Crack growth detection and estimation of depth by monitoring acoustic emission activity Article

cited authors

  • Li, C; Carballo, R; Kohlert, P; Davis, RH; Trujillo, M; Levy, C; Tansel, IN

abstract

  • Non-Destructive Test (NDT) methods have been used extensively to detect problems before actual failures occur. The monitoring of Acoustic Emission (AE) signals is proposed in this study to detect cracks in beams and to estimate crack depth. Neural networks are used for detection of crack growth and estimation of crack depth. The proposed system excites the free end of an aluminum cantilever beam by using a piezoelectric actuator with a low frequency (40 Hz square wave) signal. The Acoustic Emission (AE) activity is monitored from the fixed end and pre-processed by using a SWAN 3000 system. The existence of a crack is detected by using Adaptive Resonance Theory (ART2) unsupervised neural networks and the crack depth is estimated with Backpropagation-type supervised neural networks.

publication date

  • December 1, 1994

start page

  • 909

end page

  • 914

volume

  • 4