中文摘要：

挑选梯度算法是时域电磁反演过程的重要步骤之一,选择是否恰当直接关系到反演性能。本文精选了PRP、FR、DY、HS、LS等五种常用的共轭梯度算法,共同处理两个典型的一维问题,分别反演光滑型、非光滑型的电参数（相对介电常数和电导率）分布。仿真结果的对比表明,PRP共轭梯度法的总体性能最优。

英文摘要：

Selecting one from many conjugate gradient（ CG） algorithms is one important step for solving time-domain electromagnetic（ EM） inversion problems,therefore,whether it is appropriate or not is directly related to properties of the inversion. In this paper,five CG algorithms,namely PRP（ Polak-Ribière-Polyak）,FR（ Fletcher-Reeves）,DY（ Dai-Yuan）,HS（ HestenessStiefel） and LS（ Liu-Storey）,are selected to simultaneously reconstruct the spatial distribution of the electrical permittivity and electric conductivity from two typical one-dimensional（ 1-D） problems,for smooth type and rough type,respectively. Comparison of the simulated results indicates that the property from the PRP CG algorithm is the best one.