The Damped PSO Algorithm and its Application for Magnetotelluric Sounding Data Inversion
The Magnetotelluric inversion plays an important role in MT. Nowadays, the methods based on the layered model are used most widely. Such as the gradient method, Gaussian method, Marquardt method and generalized inverse matrix method, Bostick inversion, continuous medium inversion method, PWEI and so on. As we know that the traditional inversion methods based on the principle of minimum variance depend on the initial model and are easily trapped in a local minimum. Thats because that these inversion methods deal with the nonlinear problems by linear methods. The use of fully nonlinear inversion methods of magnetotelluric data processing has great feasibility and necessity. In this paper, we try to use a new PSO algorithm called Damped PSO Algorithm in the magnetotelluric data inversion and interpretation. We introduce the basic principles and steps of the PSO algorithm at first. And then took the numerical test of PSO algorithm. On this basis, we write a procedure of PSO for one-dimensional magnetotelluric inversion. We use this algorithm in the inversion of the theoretical data of several typical one-dimensional horizontal models (D-type and KH-type) with random noise of different levels, and compared to the inversion results of Monte Carlo method and Simulated Annealing method. Then we use it for an obverted data inversion, the result shows very well. In Abstract, PSO algorithm has strong optimization capabilities and anti-noise capability. It can be used for the initial inversion of the theoretical data and the observed data.
Particle swarm optimization (PSO) Nonlinear Magnetotelluric(MT) Inversion
XIAO Min SHI Xueming FAN Jianke ZHANG Xuhui YANG Guoshi
Institute of Geophysics and Geometics, China University of Geosciences, Wuhan, China
国际会议
The 19th International Workshop on Electromagnetic Induction in the Earth(第十九届国际地球电磁感应学术研讨会)
北京
英文
670-675
2008-10-23（万方平台首次上网日期，不代表论文的发表时间）