Research On Polynomial Equivalent Source Inversion Method

Magnetic data can’t be meshed for furhter processing,because it will bring error.And in the work of conversion calculating,the error will still exist, and be larger.Therefore,the magnetic data are caculated derectly. Generally,the magnetic data are not at a certain plane, so they can’t be processed in the frequency domain by fast Fourier transform, only be processed in the spatial domain, then, the equivalent-source method is a good choice.By following the classical approach of the equivalent-source inversion,the physical property of each equivalent source is estimated by solving a linear inversion subject to fitting a discrect set of potential-field observations.The real complex geological source is replaced by the equivalent source,and then you can perform the transform,such as, continuation,polarization, derivation.The main work and achievements in my graduation design are as follows:(1) We have researched on the classical method of the equivalent-source method, written programs and used model data to test the quality of the mathod.And we have analyzed the deficiencies of classical methods, such as a lot of computational time,a number of solutions, large memory space and so on.(2)To sovle the above problems,we have researched a method—polynomial equivalent layer inversion method, written programs and used model data to test the quality of the mathod.When the geological body is very complex, if the physical property of each equivalent source is expressed by only one polynomial, even if improved the polynomial order,we can’t achieve good inversion results. Then, we have proposed a method of piecewise polynomial inversion.(3) Piecewise polynomial inversion method makes the inversion results discontinuous. In order to solve this problem, we have designed overlapping of the sliding data window,and taken different inversion order of each segment.The method can improve the inversion accuracy.(4)The new mathod is a good method,it can improve the calculation efficiency,ruduce the number of parameters and reduce the memory consumption.