The Inversion Of Frequency Domain Soil Parameters Of Horizontally Multilayered Earth Based On Global Optimization Algorithms

A reliable soil model is the basics of calculating potential distributions of grounding grids.Generally,soil where grounding grids are buried is inhomogeneous,which means that soil electrical parameters exists difference in vertical or horizontal direction.Horizontally multilayered earth model is the most frequently used model to simulate soil electrical properties.The main work of this study is to determine soil parameters of soil layers on the basis of measurement data.The estimated model parameters include soil resistivity,relative permittivity and thickness of each soil layer.The previous studies done by other researcher show that soil parameters are frequency-dependent,which means values of soil parameters at different frequencies are different.But,thickness of soil layers is considered as frequency-independent.Therefore,frequency domain soil parameters are needed to establish a reliable soil model in frequency domain.There are two patterns in building frequency domain soil model.The first pattern is called as the direct method.For the direct method,soil parameters at different frequencies are inverted by the interpretation of spectral induced polarization data.The second pattern is called as indirect method.In that pattern,frequency domain soil parameters are described by Cole-Cole model,and the model parameters of Cole-Cole model are determined by interpreting spectral induced polarization data.Then,the soil parameters at different frequencies can be calculated by combining inversion results and the expression of Cole-Cole model.The main work of this study is as follows:Theoretical formula of apparent resistivity is derived from Green’s function with considering quasi-static field theory and dynamic state field theory.The application of complex image method avoid the calculation of numerical integration.A new electrode configuration method is proposed.The new measurement method can reduce the electromagnetic coupling effects between current lead and voltage lead.Four frequently used global optimization algorithms,including simulated annealing,genetic algorithm,particle swarm optimization and ant colony optimization,are applied to solve the optimization problem of model parameters.In order to improve the computational efficiency of inversion algorithm,parallel computing is introduced into inversion program.There are two different application way for parallel computing.For the direct method,parallel computing is applied to the inversion of each frequency point,so the inversion of each frequency point can be carried out simultaneously.For the indirect method,parallel computing is applied to evaluation part in optimization algorithms.However,this pattern can only be used in genetic algorithm,particle swarm optimization and ant colony optimization.For simulated annealing,due to its potential serial characteristic,simulated annealing can not be parallelized.Two inversion patterns are validated.By comparing the theoretical results calculated with inversion results with measured results,the accuracy and reliability of the developed method are verified.