Signal processing methods for materials defects detection

Ultrasonic Pulse Echo Technique has been used successfully as a practical modality for non-destructive testing of material to localize defects which cause critical infrastructure problems. In this paper, the pulse echo technique is used to characterize material and to localize the defects on it using ultrasound and three signal processing methods: Hilbert transform, cross-correlation function and wavelet transform. The performances of these methods are tested on synthetic signals in order to generalize the application of these methods for the detection of interfaces and therefore defects if they exist in real signals of materials. The simulation results show that the application of the signal processing methods allows the detection of interfaces and hence the location of faults and if necessary concrete characterization. Indeed, the Hilbert transform allows detection of the envelope signal locating and met interfaces. However, this approach ignores the sign gradients encountered because it considers only the absolute value. The cross-correlation function solves this problem of sign and gives a good location interfaces. The use of wavelet transform solves the problems encountered by the two other methods. Indeed, it allows a good location interfaces and therefore better detection of defects.