Experimental Ultrasonic NDT Signals of Cement paste and mortar Based on Time Frequency Analysis

The practical difficulty of extracting information to characterize materials led to use different methods of signal processing. Real ultrasonic signals used in this paper are collected from prismatic samples of paste cement without defects and mortar material with synthetic defect prepared in the NDT laboratory of Jijel University. The signals are treated by four techniques adapted for nonstationary signals, and based on energy distribution in time-frequency plan; namely: Continuous Wavelet Transform (CWT), Wigner Ville (WVD), Pseudo Wigner Ville (PWVD) and ChoiWilliams (CWD) distributions. Energy distribution in time-frequency plan allows an easy measurement of the time of flight of various echoes, and thereafter calculation of longitudinal velocity and localization of the defects. Advantages and limits of each method are as follows: WVD achieves good resolution of interfaces; however, its capacity remains limited by the appearance of non-desirable terms which may limit readings of results. PWVD avoids the phenomenon of interferences and thus appreciably improves the SNR. While it presents the problem of smoothing window's choice. CWD allows exactly extraction and clearly representation of the various components of the signal in time and frequency. Thus, improve the SNR. The experimental results show that CWT, CWD and PWVD achieve relatively good performances compared to the spectrogram and WVD. This demonstrates the effectiveness of the considered methods for experimental ultrasonic echoes for materials characterization and defect localization. However, it requires more computation time.