论文部分内容阅读
在混凝土超声无损检测中,采用CT层析成像的方法,能够呈现出混凝土内部的结构特征,更直观显示缺陷的位置、大小及类型。针对常规LSQR(Least Squares QR-factorization)迭代算法在数据误差较大时仍会发散的缺点,为了提高LSQR算法的抗噪能力,在LSQR算法迭代求解过程中施加一定的阻尼,控制减小奇异值对解的影响,这就是阻尼LSQR(Damping Least Squares QR-factorization)算法。计算机仿真结果及混凝土试件试验结果表明:阻尼LSQR算法比LSQR算法具有更好的CT重建效果和消噪能力,提高了计算的精度和图像重建质量。
In the ultrasonic non-destructive testing of concrete, the method of CT tomography can reveal the internal structural features of concrete and more directly display the location, size and type of defects. In order to improve the anti-noise ability of LSQR algorithm, a certain amount of damping is applied during the iterative process of LSQR algorithm to reduce the singular value The effect on the solution, which is damping LSQR (Damping Least Squares QR-factorization) algorithm. Computer simulation results and test results of concrete specimens show that the damping LSQR algorithm has better CT reconstruction effect and denoising ability than LSQR algorithm and improves the accuracy of calculation and image reconstruction quality.