Study Of Improved Method Of Quality Of Acoustic Traveltime Tomography

Acoustic tomography is a non-destructive testing technology being widely used in amount of domains like medical science, industrial inspection geological prospecting and so on. The image is accurate, visualized, adjustable and non-destructive for the object, so it is extremely valuable to research the sonic tomography. With high accuracy, stability and computational efficiency, the damped least squared method is one of the most common linear inversion method. However, the inversion result is seriously affected by the position of geometric raypath and relies heavily on the initial value. In this paper, we demonstrate a method to improve that imperfection and enhance the image quality by amending the eigenvectors of model space. As follows:Firstly, this paper introduces the application value and significance of sonic tomography and image quality improvement. And the development and some major approaches of image quality improvement are introduced in this paper. In addition, this paper also introduces the basal principle of sonic tomography, such as some mathematical and physical methods, the realization processes, including data acquisition, data processing and result assessment, and some major approaches for each process.Secondly, with the large-scale sparse equations after linearization of inversion problem, this paper determines whether the equations is ill-condition by analyzing the condition number and singular value of coefficient matrix. And the damped least squared method is a frequently-used inversion method, which utilizes a suitable damping factor to reduce the scale of condition number of normal matrix and resolve the ill-condition on the liner system. In addition, this paper points out that the inversion result is seriously affected by the position of geometric raypath and relies heavily on the initial value, which is the main disadvantage of damped least squared method, and the reason of the affection is researched in this paper by analyzing the eigenvectors of model space. Finally, based on the significance of the first eigenvector of model space, a new optimization method is put forward in this paper to enhance the image quality of damped least squared method by amending the first eigenvector. The optimization method is compared with damped least squared method without correction in both homogeneous, stratified and block-structured model. It is demonstrated at last that the optimization method proposed in this paper is effective on reducing the impact of geometric raypath and improves the imaging quality greatly.