Study Of Magnetotelluric Nonlinear Inversion Based On Artificial Bee Colony Algorithm

Magnetotelluric(MT)method is a geophysical exploration method to study the distribution law of resistivity of underground rock strata by using the differential characteristics of natural electromagnetic field propagation and distribution in underground.This method has been widely used in the exploration of mineral resources,geothermal,water resources and oil and gas resources.However,with the deepening of theoretical research and the popularization of practical application,there are many new problems that need to be further solved and improved.The study of magnetotelluric inversion is a key research direction at present.Magnetotelluric sounding inversion mainly solves the minimum value by constructing an appropriate error objective function,and its mathematical essence is the optimization problem.For this reason,many geophysicists have proposed various inversion methods.Such as Bostick inversion,continuum inversion,Gauss-Newton method based on variance principle,Marquette method,generalized inverse inversion and so on.But the traditional inversion method based on the principle of minimum variance,because the inversion results always depend on the initial given model,is easy to fall into local minimum.In this paper,a nonlinear inversion method of magnetotelluric(MT)based on artificial bee colony algorithm is proposed to improve the inversion accuracy.Artificial bee colony algorithm is a stochastic optimization algorithm based on intelligent search behavior of bee colony proposed by Professor Karaboga of Ergis University in Turkey.A large number of experiments show that the ABC algorithm has the advantages of fast convergence speed,high quality of non-inferior solutions and good robustness in optimization.It is very suitable for solving various numerical optimization problems.Firstly,the basic principle of magnetotelluric sounding method is introduced in detail.Starting from Maxwell's equations,the ground impedance recurrence formula of one-dimensional N-layer medium magnetotelluric field is deduced.The forward program is compiled by using MATLAB.The forward calculation of two-layer,threelayer and four-layer geological models is carried out,and the calculation results are analyzed.The foundation for inversion is laid.Subsequently,the basic framework of nonlinear inversion based on artificial bee colony algorithm is constructed,and the objective function of artificial bee colony algorithm is improved according to the characteristics of magnetotelluric sounding curve and the precision and electrical characteristics of inversion.The number is solved by a nonlinear method.The feasibility and practicability of magnetotelluric nonlinear inversion method based on artificial bee colony algorithm are discussed.