中文摘要：

从MT信号中提取激电信息的研究对深部矿产资源和油气资源的开发具有重要的现实意义.在激电信息提取方法的研究中,过去的理论多以线性反演方法为主,主要存在三个问题：（1）依赖初始模型,（2）容易陷入局部极值,（3）多解性严重.本文通过以下方法解决上述问题：（1）采用差分进化算法对MT信号中的激电信息以反演的方式进行提取,该方法受初始模型影响小,不易陷入局部极值.（2）在差分进化算法的适应度目标函数中引入最小构造约束,并提出对极化率和电阻率分别约束来提高反演结果的稳定性.对极化层处于不同位置的不同地电类型的反演仿真结果表明,本文算法能有效获得电阻率和激电参数的光滑模型,反演结果具有较高的稳定一致性和准确性.加入噪声测试后的实验结果表明,本文算法对高斯白噪声有较强的鲁棒性.与微粒群优化算法（PSO）比较结果表明,本文算法具有更为优越的收敛速度,能够获得更好的反演效果.上述结果表明,用非线性方法提取MT信号激电信息具有积极的意义.

英文摘要：

The study of the IP information extraction from magnetotelluric sounding data is of great and practical significance to the exploitation of deep mineral resources and oil and gas resources.The linear inversion method is given priority to on previous research of IP information extraction method,which has three main problems：（1）Depends on the initial model,（2）Easily falls into the local minimum,（3）Serious non-uniqueness of solutions.The following methods are used to solve the above problems in this paper：（1） Differential Evolution algorithm is used to extract the IP information in the form of inversion.The inversion results are less affected by the initial model and are not easy to get into the local extremum.（2）Minimum structure is employed to constrain the fitness function of DE algorithm so as to solve the problem of multisolution in inversion.And we propose that restriction should be imposed respectively on polarizability and resistivity in order to improve the stability of the inversion results.The inversion simulation results of polarization layer in different stratum of various geoelectric model show that the smooth models of resistivity and IP parameters can be obtained by the proposed algorithm,the results of which have higher stability and accuracy.The experiment results added with noise indicate that this method is robust to Gaussian white noise.And when compared with PSO algorithm,this algorithm has better convergence rate and inversion effects.The above results show that nonlinear method plays a positive role in the IP information extraction of MT signal.