The Study Of Non-linear Inversion Theory Of DC Resistivity Sounding

DC resistivity sounding, as a sophisticated geophysical methods, how to improve theaccuracy of its information retrieval, inhibition of multiple solutions, and improve theresolution of the inversion results has been the majority of workers continue to exploregeophysical problems.The inversion of resistivity sounding is a multi-parameter, multi-extreme and non-linear,and its essence is in the observation data and forward model parameter space to find theoptimal functional solution. previous non-linear inversion method is the problem more linear,spent high order terms in the objective function optimization iteration, the consequences ofthis inversion is not only heavily dependent on the selection of the initial model, and easilyfalling into local optimal solution, easy to overlook some details, and because it is the matrixoperation, in dealing with large amounts of data often makes the solution process will be slow,and even affect the stability of solution.In recent years, non-linear inversion is explore mainly directions of geophysical workers,the basis of a random process, From the proposed Health, quasi-physical point of view,traverse the model space, or heuristic search, Unlike single-point iterative linear inversion, thepopulation units to multi-point search, while avoiding the partial derivative of the objectivefunction of the complex process and can quickly achieve the objective of global optimization.The article from the sounding of the theoretical analysis, carefully studied the forwardsounding theoretical model and related resistivity forward formula, On this basis, theinversion for sounding more outstanding issues into the current field of swarm intelligencealgorithms for nonlinear-differential evolution algorithm, the algorithm to solve nonlinearproblems has unparalleled advantages, the main difference is the use of groups within thepopulation vector direction and the exchange of information between individuals does notdepend on gradient information heuristic global search, based on complete control theprinciple,basic structure, algorithm, other related features and content expansion mode ofDifferential Evolution Algorithm, For the first time established a retrieval model based ondifferential evolution algorithm, and then test the theoretical model by comparing theparameters of the differential evolution algorithm-depth study of selected issues, Identified aset of reasonable parameters of the inversion, and through the platform of Matlab achieve theautomatic inversion of fitting function of resistivity sounding curve. finally, the program willprepare the actual promotion of information typical of known holes drilled beside the sounding data inversion, through the comparative study of the inversion accuracy andgeneralization ability, and by the nonlinear inversion methods and the existing inversionmethods were compared, realize the model optimization, and then to assess its generalizationability.With model test results and the actual promotion of the differential evolution algorithm,both from the increased speed of retrieval accuracy and speed up the inversion, etc., thedifferential evolution algorithm has obvious advantages than the previous linear inversion andother non-linear inversion method, And its robustness, the calculation takes less memoryresources. the results of this project opened up a non-linear geophysical inversion theory newarea, we can applied it to two-dimensional or three-dimensional inversion of resistivitysounding theories in further.