Rayleigh Wave Dispersion Curve Inversion Combining With ABC And DLS

Rayleigh wave is a kind of seismic wave formed by the mutual interference of P wave and S wave and propagating along the free interface.Because Rayleigh wave has dispersion characteristics,velocity is a function of frequency,as long as the formation velocity changes slightly,its frequency will also change.Therefore,through the inversion of the Rayleigh wave dispersion curve,the thickness,shear wave velocity and Q value related to the formation structure and physical properties can be obtained.At present,this technology has been widely used in the detection of the internal structure of the earth,especially to solve the geological problems in the fields of near-surface engineering and environment.Inversion of Rayleigh wave dispersion curve is one of the important links in Rayleigh wave exploration.It is a multi-parameter and multi-extreme nonlinear iterative optimization process.At present,there are two major branches: linear inversion and global nonlinear inversion.Linear inversion has the advantages of high computational efficiency and less memory,but it has higher requirements for the initial model.Global nonlinear inversion has a low degree of dependence on the initial model,but it is time-consuming and low in inversion accuracy.For this reason,this paper proposes a two-step inversion strategy of combined artificial bee colony algorithm(ABC)and damped least squares(DLS)to invert the Rayleigh wave dispersion curve.First,on the basis of prior geological information,the nectar source is initialized in a larger model scale;Then,the three types of bees(employed foragers,unemployed foragers,and scouts)are transformed into each other and several iterations of search optimization are performed to converge the target to the optimum The solution is near to obtain the initial shear wave velocity model;Finally,the DLS with fast convergence speed and strong stability is used for local optimization and inversion to obtain the final shear wave velocity model.Due to the disorderly random update of nectar sources and the blind search of bees in the ABC algorithm,it is difficult to give full play to the global search performance of the algorithm.This paper uses the global optimal solution guidance and cross factor constraints to form a cross global bee colony algorithm(CGABC)with higher search efficiency.In the later stage of CGABC optimization,the non-linear stochastic development honey source was replaced with DLS linear optimization,which further improved the calculation efficiency and inversion accuracy.The correctness of the method is verified by multiple sets of model calculations and further applied to a set of actual seismic data processing.The research results show that the inversion strategy in this paper not only effectively reduces the dependence of linear inversion on the initial model,saves the calculation cost of nonlinear inversion,but also significantly enhances the stability of inversion and effectively improves the accuracy of inversion.