Rayleigh wave dispersion curve inversion method for seismic surface wave exploration

Abstract

The invention belongs to the field of near-surface seismic surface wave exploration, and particularly discloses a Rayleigh wave dispersion curve accurate and fast inversion method for seismic surfacewave exploration. The method includes converting a seismic record from an x-t domain to an f-k domain to obtain f-voR through two-dimensional Fourier transformation of the seismic record; according tothe f-voR, randomly generating an initial population; calculating phi (Pj), converting the phi (Pj) into f (Pj), and recording that the initial iteration number of inversion is 1; according to the f(Pj), preforming selection operation and cross operation; based on a cross population, selecting the top 10% individuals of f (Pj), performing local linear optimization, and acquiring a new populationPj (x) through the fastest descent; and based on the new population, repeating the above steps until T = T+1 reaches the preset iteration number, and selecting the individual with the largest fitnessvalue as an inversion result according to the f (Pj). The method can improve the applicability, rapidity and accuracy of Rayleigh wave exploration in detecting near-surface stratum structures.

Description

technical field [0001] The invention belongs to the field of near-surface seismic surface wave exploration, in particular to an accurate and rapid inversion method for Rayleigh wave dispersion curves used in seismic surface wave exploration. Background technique [0002] Rayleigh wave exploration uses its dispersion characteristics in layered media to construct the formation S-wave velocity structure model by extracting and inverting the dispersion curve, which is fast, convenient, and non-invasive, especially in very high shallow layers. advantages such as resolution. Among them, the dispersion curve inversion is the key link connecting the entire exploration process. Since it is a multi-parameter, multi-extreme, multi-modal, highly nonlinear iterative optimization process, the conventional damped least squares (DLS) inversion is because Its strict initial model requirements are often not applicable. Although the improved equal-thickness and thin-layer DLS application is r...

AI Technical Summary

Problems solved by technology

Among them, the dispersion curve inversion is the key link connecting the entire exploration process. Since it is a multi-parameter, multi-extreme, multi-modal, highly nonlinear iterative optimization process, the conventional damped least squares (DLS) inversion is because Its strict initial model requirements are often not applicable. Although the improved equal-thickness and thin-layer DLS application is relatively common, the division of the formation into several equal-thickness and thin layers has caused a large increase in inversion parameters, and the problem of multi-solution is also very large. Outstanding; the nonlinear inversion results based on global optimization are easier to match with the actual geophysical model, but to obtain accurate inversion results and high inversion accuracy, traditional genetic algorithms must rely on a huge selection The sharp increase in the number of groups and training times, the amount of calculations, and the memory usage lead to a very low inversion efficiency; the DLS inversion after the initial optimization of GA overcomes the disadvantages that it is difficult to apply effectively, and can suppress the inversion of multiple solutions to a certain extent However, it is still difficult to obtain the global optimal solution when dealing with more complex multi-solution problems caused by the existence of high / low-speed interlayers and thin layers.

Images