Constrained Sparse Spike Seismic Inversion With Regluarzing And Its Application

Impedance is a key par ameter of the spatial distribution of the reservoir condition and prediction, it can reflect reservoir direc tly. So the im pedance inversion technique has been done intensive research by a large number of researchers. This paper proposed a regularizing method with sparse spike inve rsion to solve the pr oblems of impedance inversion, which has non-unique solutions and the result is not high ly accuracy. The regularizing method can turn the ill-posed inversion problem into good-posed one, and thus the solutions can be more stable. Meanwhile, the process of impedance inversion is complicated, involving m any factors. So, we analyze the related factors of seism ic inversion by the theory test.Constrained sparse spike inversion has the advantages: less dependent on the initial model, including low frequency, and acquiring the broadband reflection coefficients by the less logs and so on. So it has been wide ly used in actual production. Focusing on this method, and com bining with the optim ization theory, the prim arily work and achievements of the paper are as follows:(1)Study on the basic theory of constraine d sparse spike invers ion, including the basic hypothesis if constraine d sparse spike inve rsion, the sparse representation of reflection coefficient sequence, the relationship between im pedance and reflection coefficient, the objective function and lo w-frequency compensation, and optimization algorithms, etc.(2)Study on the regularization m ethod, especially T ikhonov regularization. Combining the research status, we impr oved it by s mooth operator and regularization parameter.(3)The influencing factors of seism ic inversion are complexity and diversity. We analyzes on the m ain affecting factors. And we should pay m ore attention to these factors and insure each links in the process of seismic inversion.(4)Core algorithms with C++ were re alized and software m odules of the constrained sparse spike inversion on QT development platform based on Windows/Unix operating system. And we pr ove the co rrectness and practicability of methods by practical testing data.