91 Application and Study on QR-Based Least-Squares Method Algorithm in Concrete Ultrasonic Computerized Tomography

The structural characteristic is given by computerized tomography in non-destructive testing (NDT) of concrete. The position, size and style of inner defects can be shown. In view of the large date error, Least Squares QR-factorization (LSQR) algorithm will diverge. In order to improve the anti-noise capability of LSQR algorithm, inhibit the data divergence in the iterative process and reduce the impact of singular value on the solution, iterative process can be imposed a certain damping. This is damping Least Squares QR-factorization algorithm (DLSQR). Computer simulation and concrete model of ultrasonic computerized tomography show that the DLSQR is more effective to improve the accuracy of computation and the quality of image reconstruction on concrete than the LSQR.