中文摘要：

三维电阻率探测在实际工程中日益受到重视，但存在着深部分辨率低、反演效率低等关键问题，严重制约了三维电阻率探测技术的应用和推广。针对深部分辨率低的问题，提出了一种随三维模型网格尺寸自适应调整的加权光滑约束，改善了深部网格的电阻率差异容许程度，实现了对不同深度网格约束的差异化加权处理，有效的提高了深部反演的分辨率和成像效果。针对反演计算耗时长、效率低的问题，基于预条件共轭梯度法求解快速稳定的优势，提出了三维电阻率快速稳定反演成像算法。在该算法中，将雅可比迭代中的对角阵作为预条件矩阵，其具有求逆方便、无需内存空间的特点，显著加快了收敛速度。最后，利用合成算例和隧道导水裂隙探测的工程实例验证了上述反演方法的可行性与有效性，表明借助于自适应调整加权光滑约束和预条件共轭梯度算法，有效的提高了深部分辨率和计算效率，显著改善了反演效果。

英文摘要：

Three-dimensional resistivity detection is increasingly concerned in actual projects. However, its application and extension is excessively restricted because of some key problems such as low deep resolution and low inversion efficiency. For the problem of low deep resolution, an adaptive-weighted smooth constraint is proposed, which is adjusted adaptively to the mesh size in a three-dimensional model. It improves resistivity difference tolerance of deep grids and achieves difference weighted processing of grid constraints at different depths. With the adaptive-weighted smooth constraint, the deep resolution and inversion imaging quality are improved effectively. For the problem of time-consuming and low inversion efficiency in inversion, a rapid and stable method of 3D resistivity inversion and imaging based on preconditioned conjugate gradient （PCG） algorithm is proposed. The diagonal matrix in Jacobi iteration is used as the preconditioned matrix for speeding up the convergence speed significantly. The inversion of the preconditioned matrix is convenient to be solved and doesn＇t occupy memory spacing. At last, the inversion method is applied in a synthetic example and water flowing fracture detection in tunnel engineering for checking its feasibility and effectiveness. The results show that the deep resolution, calculation efficiency and inversion quality of 3D resistivity detection are improved effectively by using the adaptive-weighted smooth constraint and the PCG algorithm.