Research On Multimodal Rayleigh Wave Dispersion Curve Inversion Based On Improved Nonlinear Algorithms

Rayleigh wave exploration is a very convenient,economic,reliable and effective exploration method,which has been widely used in geotechnical,engineering seismic and shallow surface survey.By inversion of Rayleigh wave dispersion curve,a series of formation evaluation parameters can be obtained effectively,such as shear wave velocity,formation thickness,Poisson's ratio and foundation bearing capacity,etc.Inversion of Rayleigh wave dispersion curve is a key step in Rayleigh wave exploration data processing.At present,the inversion methods of Rayleigh wave dispersion curve are mainly linear and non-linear.With the development of computer technology,the non-linear inversion algorithm are gradually taken seriously,and the unique advantages of them are gradually revealed,For example,they do not depend on the initial model and they are easy to implement,moreover they are also suitable for solving the inversion problem of Rayleigh wave dispersion curve which had high nonlinearity,multi-extremum and multi-parameter.Therefore,it is becoming more and more important to study the application of non-linear algorithm in Rayleigh wave inversion.Genetic algorithm,particle swarm optimization and artificial bee colony algorithm are relatively mature,which can solve the problem of Rayleigh wave dispersion curve inversion better.However,due to the slow convergence speed,premature convergence and poor convergence accuracy of the three algorithms,the efficiency of their direct application to Rayleigh wave inversion is extremely low.At the same time,the dispersion curve of higher mode is more sensitive to shear wave velocity and formation thickness than that of basic mode.Therefore,more accurate shear wave velocity distribution information can be obtained by inversion of multi-mode dispersion curve.In order to solve the above problems and combine the characteristics of Rayleigh wave inversion,three kinds of non-linear algorithms were improved as follows:(1)Aiming at the slow convergence speed of genetic algorithm,a three-neighborhood competitive search strategy is proposed after crossover and mutation,and an improved LGA algorithm is obtained to accelerate the convergence speed of the algorithm.(2)By Improving the inertia weight coefficient of particle swarm optimization(PSO)in speed updating by using tangent function adjustment strategy,and adding the cross-mutation process of genetic algorithm,the improved GAPSO algorithm is obtained.(3)In order to improve the local exploring ability of the artificial bee colony algorithm,the search strategy of the artificial bee colony algorithm is improved in the hiring bee search stage,and the search mode and tracking mode of the cat colony algorithm are added according to the quality of the honey source after the following bee stage,then the improved HABC algorithm is obtained.After all improvements have been completed,three improved non-linear algorithms are used to inverse the third-order dispersion curve with and without noise for three typical geological models: increasing velocity model,low-velocity interlayer model and high-velocity interlayer model.In addition,this paper also implemented the inverse analysis of three improved algorithms after extracting the dispersion curve from the measured data.The results obtained from the above calculation show that the three improved non-linear algorithms can effectively and steadily invert Rayleigh wave dispersion curves,and obtain more accurate and reliable inversion results.At the same time,the improved HABC algorithm and GAPSO algorithm have greatly improved the convergence speed and accuracy.The improved LGA algorithm inherits the inversion accuracy of the original algorithm while speeding up the convergence speed.