Rayleigh Wave Dispersion Curve Inversion Combining With Adaptive GA And DSL

The Multiple Transient Surface Wave method is one of the most widely used and most effective methods in the field of near-surface exploration such as geotechnical investigation,geological disaster investigation,etc.It utilizes the dispersion characteristics of surface wave propagation in layered media.Extracting and inverting dispersion curves are used to construct the stratigraphic model of S-wave velocity,and to further characterize the media's attribute information in different depth ranges,such as density,Poisson's ratio,and quality factor.It has the advantages of fast,convenient,non-invasive and shallow resolution,etc.Rayleigh wave dispersion curve inversion is one of the most important contents of surface wave data processing,and it has the key role of linking up and down.For its inversion processing,there are two linear and nonlinear theoretical systems.The least-squares damping method(DLS)is the most mature and commonly used method for linear inversion,but it depends heavily on the established initial model,otherwise it easily falls into a local minimum,and even leads to wrong inversion conclusions;The nonlinear inversion is represented by the genetic algorithm(GA),and it is theoretically easier to match with the actual geophysical model,especially for inversion processing under complex near-surface geological conditions.However,the computational complexity and the large memory footprint cannot be ignored.In addition,the convergence efficiency by mutation operations with low probability of occurrence and uncontrollable direction produce “excellent genes” drive the entire population toward the global optimal solution is extremely low especially in the late inversion.And it cannot be completely avoided that the global optimal solution cannot be obtained due to premature convergence.In the past,the above two method systems have been independently studied for inversion processing.The problem that the linear inversion seriously depends on the initial model,it is easy to fall into the local optimum,and the instability of the nonlinear inversion process and the inefficiency cannot be fundamentally eliminated.In order to solve the above problems,this paper carried out a study of “Rayleigh wave dispersion curve inversion combining with GA and DSL”.The research results show that this method can suppress the multi-solution problem of Rayleigh wave dispersion curve inversion obviously,improve the accuracy of model parameters,and take into account the timeliness and accuracy of inversion.Providing method support for near-surface fine structure detection and surface wave reuse in oil and gas seismic records.The main research results obtained in the paper are as follows:(1)Based on the traditional dichotomy method for solving the Rayleigh wave dispersion function,an algorithm is proposed which has a fixed frequency and multiple phase-velocity directions to find the dispersion.This effectively avoids the problems of leak the root and misjudgment of patterns with conventional algorithm when calculating the high-order dispersion curve,and further improves the accuracy of multi-mode dispersion curve forward modeling.(2)The two-step joint inversion method of GA and DLS can effectively get rid of the high dependence of the conventional DLS inversion on the initial model,and can be used to solve the problem that the DLS algorithm is difficult to apply due to lack of prior geological information or incomplete data in actual data processing.The two-step joint inversion algorithm can achieve accurate inversion of the parameters of the conventional bedding model.(3)Adaptive GA and DLS nested joint inversion method by adding the adaptive “genetic operator” to the selection,crossover and variogram of classical GA inversion has effectively accelerated the inversion process.At the same time,DLS is used to guide the dominant individuals(ie,multiple local extremums)to quickly and steadily advance toward the global optimal solution.The method is more suitable for the inversion of complex geological models with high,low velocity interlayers,thin layers,etc.It can effectively suppress the inversion of multiple solutions and improve the search efficiency and inversion accuracy of the global optimal solution.(4)The large distance between single-shot seismic traces in oil and gas exploration results in a narrower surface wave frequency band and high-band energy shortage,making conventional inversion methods more problematic due to lack of more effective information constraints.Based on effectively suppresses the inversion of multiple solutions to dispersive curves with joint inversion method of adaptive GA and DLS method in this paper,and its trial processing shows that this method can perform this problom obove and can achieve good application results.The acquired geological parameters can be directly used for the processing and interpretation of conventional seismic data.