中文摘要：

本文通过分析各向异性扩散（Anisotropic Diffusion AD）在信号的去噪与增强中对特征的保护特性，及各向异性扩散与Robust估计之间的关系，把各向异性扩散引入到阻抗估计中，提出了基于各向异性扩散的Robust阻抗估计方法（ADR）．模拟结果表明该方法对噪声或“飞点”具有明显的压制作用，与最小二乘法（LS）及常规的Robust估计方法相比具有更好的稳定性和精度．当噪声方差从0增加到0．8，ADR估计的相对误差最小为4．0％，最大仅为4．9%，误差均值为4．5％；Robust估计相对误差最小4．1％，最大6．2％，平均为4．9％，LS方法平均相对误差则达到18．4％．Robust估计收敛过程受阈值的影响较大，ADR收敛速度虽然受到扩散步长与阚值的影响，但收敛值与阈值选择无关．这在实际应用中有着重要意义．

英文摘要：

It is analyzed that the preserve character of Anisotropic Diffusion （AD） to features in signal denoising and enhancing, and the relations of Anisotropic Diffusion（AD） and Robust Estimation in the paper. Anisotropic Diffusion is introduced to impedance estimation, and Robust impedance estimation method（ADR） based AD is proposed. The results of numerical simulation show that the ADR can remove noise and reject ＂outliers＂ effectively, and it is better stability and accuracy compared with least square （LS）and traditional Robust estimation. When noise variance increases from 0 to 0.8, minimum, maximum and average of ADR estimation relative error are 4.0%, 4.9% and 4.5%respectively; those of Robust Estimation＇s are 4. 1% ,6.2 and 4.9% respectively; and average relative error of LS is up to 18.4%. The convergence process of Robust Estimation is larger affected by threshold, although the convergence velocity of ADR is affected by threshold and diffusing step, the convergent value has nothing to select threshold. That is very important for impedance estimation in the field data.