Research Of Improved PSO Algorithm For Electrical Resistance Tomography Two-dimensional Nonlinear Inversion

As a method of the traditional geophysical exploration,electrical prospecting has been widely applied in geological examination,mining exploration,ground water and oil exploration etc. Electrical Resistance Tomography(ERT for short) is a kind of typical electrical prospecting method which has become one of the topics of geophysical technology with its simple realization,convenient implementation of field source,high resolution,large penetration depth etc advantages.Thus, the research of ERT has the widespread application prospect. Because the calculation precision and effective interpretation of resistivity profile data are the important indicator of its application,ERT resistivity data inversion interpretation is the primary means of dealing with actual data.Therefore,ERT data inversion has also been the focus of extensive research at home and abroad.In terms of improved particle swarm optimization(PSO for short) nonlinear inversion of the two-dimensional simulated ERT date,this paper conducted a comprehensive and meticulous research, mainly including:A brief development of ERT and its inversion status isgiven.In the next part, paper expounded the basic principle of PSO algorithm and the related improvement. In view of the standard PSO global search later convergence slowly and easily trapped into local optimum, author proposed the improved PSO algorithm by introducing chaos search optimizer to parameters that influenced particles velocity updating, and tested with typical test functions compared to the standard PSO algorithm and the damping of PSO algorithm.The results show that improved PSO algorithm speeds up the convergence rate, improves the precision of convergence.Next,it derived a discrete type of the 2-D partial differential equation by using finite element method with unstructured triangular elements.Then,the results of wave number domain in the forward matrix is obtained.Finally,the results of the wave number domain transform into spatial domain by the Fourier inverse transformation conversion and the forward research was completed.We prepared the forward FEM numerical simulation program in Matlab 2012 b and model experiment proved that FEM forward with unstructured triangular grid used in this article was feasible,and laid the foundation for the next chapter.The next part demonstrated achievements.We illustrated2-D ERT inversion theory, derived the minimum objective function of nonlinear inversion,and inverted forward model originally designed for DC resistivity、TDIP data and FDIP complex resistivity.In order to ensure the authenticity of the simulation field operations, we added a certain amount of noise to the forward model data.Eventually, paper gave three model inversion results of DC resistivity、TDIP data and FDIP complex resistivity,compared the performance improved PSO algorithm with standard PSO algorithm and Occam algorithm.The model experiments show that the improved PSO nonlinear inversion method is not dependent on the initial model,increases the search space,and have higher inversion accuracy and not easier to be trapped into local optimum than the standard PSO.Compared with the damped PSO, it has faster convergence speed and higher performance.