Study On Strategy Of Ultrasonic CT Inversion Of Concrete

There is no doubt that it is the important basis of nondestructive test in concrete to take advantage of the traveling features of ultrasonic wave in concrete. Therefore, the first chapter of this thesis is devoted to discussing the traveling features of ultrasonic wave in concrete.Then, this thesis discusses the use of wavelet transformation in ultrasonic signal's denoising. Through simulating experiment, the wavelet denoising method appropriate for ultrasonic signal, as well as the optimum wavelet base, is found. Besides, wavelet transformation possesses a relatively strong capability of analyzing time and frequency domain locally. Analyzed by wavelet, the ultrasonic signal can be divided into many different frequency domains. A flaw factor used to characterize the flaw is proposed by analyzing the frequency domain which has most of the signal's energy. The smaller the flaw factor is, the more serious the flaw is. The experiment shows that this factor can accurately reflect the flaws'degree.Following that, this thesis proposes a qualitative analytical method of imaging concrete flaws by ultrasonic test, which is developed from the traditional sound shadow superposition law. Just as tomography, it can rebuild the images of specimens'transects, but the amount of calculation is far less, and its programming is far more simple, it can obtain the message of the inner concrete's quality from the qualitative angle. At the same time, it can provide another kind of image for the quantitative calculation of inversion. Combine these images with the ones obtained through quantitative calculation, the accuracy of the inversion can be increased.At last, this thesis discusses the application of three modern intelligent optimization algorithms in computerized tomography of ultrasonic velocity in concrete. At the same time, this thesis makes some improvement to the traditional DLSQR algorithm, and through model experiment, all the algorithms are compared, in order to find out their shortcomings and strongpoints. The experiment results show that as for the four algorithms used in this thesis, their numerical values are stable, and their results are correct. In this thesis, the DLSQR algorithm is used to solve the normal equation of augmented objective function of quadratic optimization, which makes the velocity of each grid unit restricted correspondently in order to make them get closer to the practical physics model, but still some of grid units'velocity are relatively high. As for Chaotic Simulating Annealing and Re-annealing Algorithm(CSARA), Chaotic Simulating Annealing and Genetic Algorithm(CSAGA), and Continuous Ant Colony Optimization (CACO), as the velocity values of inversion have already been restricted effectively in the process of searching optimum values, their inversion results fit the practical physics model better. Besides, among four algorithms, DLSQR, which possess advantage in calculating speed and fitting residual of traveling time, is a stable and quick inversion algorithm. Among other intelligent optimization algorithm, CSARA has the best efficiency of searching optimum, while the other two, as they relate to the intergenerational evolution calculation, the amount of calculation is relatively large, and their speed of searching optimum is relatively slow, furthermore, their calculating speed declines immediately with the enlargement of the matrix of ray trace. Between them, as CSAGA has the Chaotic Simulating Anealing Algorithm to search locally, its efficiency is higher than CACO. Therefore, among the three intelligent optimization algorithms, CSARA is recommended.Moreover, this thesis still discusses the influence of different image post-processing method, different flaw combinations in concrete and flaws of different properties on the quality of ultrasonic CT images.