Studies On Crack Identification Methodalogy For Structure Based On Mode Analysis And Stress Wave

Early damage detection is an important method to provide the safety of the performance for civil engineering structure. The topic of this paper is how to identify the crack parameters in the structure. First of all, the state and developement of the problem in the last 40～50 years are reviewed. Based on it, the four-step strategy on crack parameters identification was proposed, which include damage index; damage model; solve and optimization. A successful parameters identification method must solve the 4 itemes in the above. It is a frame for the crack parameters identification.As the following, the methods base on frequency, wavelet analysis, the beam response under the moving load and stress guided wave were studied in detail. For reliablity and inexpensivety, the methods based on natural frequency changes had the principal attraction in this field. However, the frequency is not a sensitive indicator for the local damage. As results, the author developed a method based on wavelet transform of the mode shape. The crack location could be found by the sudden changes in the wavelet coefficients. Furthermore, the defined intensity factor relating the wavelet coefficients was used to estimate the size of the crack. Contrasting the merits of the frequency-based and the wavelet-based methods, we could find that the wavelet-based method is convenient on locating and numbering cracks, while the frequency-based one is reliable on sizing cracks. An identification method combining wavelet analysis with transform matrix was developed in this paper. Firstly, the fundamental vibration mode was applied to wavelet analysis. The crack location was found by the peaks of the wavelet coefficients. Secondly, based on the identified crack locations, a simple transform matrix method requiring only the first two tested natural frequencies was used to further identify the crack depths. Even so, it is very difficult to measurement mode shape using the tradinal modal test. To solve this problem, the crack detection method based on the response of the single point at the beam under the moving load was also proposed in this paper. The response at mid-span of the beam was wavelet packet transformed (WPT). As results, the crack could be found by the abnormal signal in WPT branches and the size was estimated by a defined damage index (Dindex). And the effect of both crack location and wavelet selection on Dindex was discussed in detail.It has to vibrate the structure to obtain the mode parameters using the above methods. But, it is not easy to vibrating special structures, such as a dam, buried pipe and other no-accessible Component. These methods are possibly failed if one detecte the crack in these special structures. Crack identification based on Stress Guided Wave (SGW) is another way of crack identification. And its academic base is the propagation of the stress guided wave in the solid media with the boundary. There is no need to force the structure to vibration. But, the success in application using SGW is not easy because of the propagation of the SGW accompanying complex disperse, mode transition and attenuation. The author tried to study the propagation in the pipe, and developed two sensitive methods to identify the crack using the SGW. One was time-shift correlation analysis, whose correlation coefficients between the excitated signals and measuremneted guided wave at different time were calculated with the help of a window function. The arriving time of the echo from the damage could be found in the correlation coefficients; the other was based on wavelet transform, by which the echoes from the damage could be discerned in the wavelet coefficients.Finally, experiment studies based on piezoelectric transducers were performed, that crack parameters identification using ultrasonic guided wave. Guided wave of mode L(0,2) was successfully launched using annular piezoelectric transducers. The results showed that the frequency and the number of period of guided wave are the sensitive factors. The launched mode and output power heavily depend on the frequency. In general, more periodic guided wave induces to bigger power output but its fault is likely to aliasing between the incident wave and reflected wave for the wide time-span signal. As this experiment, the measurements are an excellent result using the 20-period and 70kHz guide wave. The dispersion curves of the group velocity and the attenuation coefficient are also obtained, and it has a good consistency with the theoretical value. To reduce the noise from the guided wave signal, both 8 points FFT Smoothing technology and wavelet multi-resolution analysis based noise reduce methods are usd to denoise the guided wave signal. The results shows that the denoise quality using wavelet multi-resolution analysis is better than the one using FFT smoothing obviously. The reflection coefficient is defined by the ratio of the peak of the echo from the crack to the peak of the incident wave. When the modify factor is 0.8, whose aim is to reduce the effection of the dispersion, it could be found that the size identification results are in good agreement with the theoretical prediction.