Research On Seismic Wavelet Extraction And Acoustic Impedance Inversion Based On Non-liner Methods

Seismic wavelet extraction and stratum acoustic impedance inversion are very important in seismic inversion. In this thesis, the author utilizes prevalent non-liner methods such as fractal theory, Genetic Algorithm (GA), Simulated Annealing (SA) to extract wavelet and inverse acoustic impedance. Some theoretical and practical data show the effects of these methods.On the seismic wavelet extraction, using the fractal characteristic of power spectrum of reflection coefficient series, combining multi-trace statistical cepstrum method based on autocorrelative theory, the author develops a method by cepstrum to extract wavelet, which based on fractal technique. The method is applicable to extract wavelet when the reflection coefficient are non-white noise random series. Theoretical model and practical data processing results show this method can extract seismic wavelet more accurately and improve seismic inversion precision. That is, the method has better effect to improve the accuracy of seismic wavelet inversion.On the acoustic impedance inversion, first, applying generalized non-liner least square method to inverse acoustic impedance, taking account of the advantage of GA and SA, the author studied a new method, which combined the three methods just mentioned, and called gradual united inversion (GUI) technique. The simulated test and practical data processing results show the GUI technique can improve the resolution of the acoustic impedance inversion, and in some extent it can restrain the multi-solution of acoustic impedance inversion.