本文对常规最小熵反褶积方法中存在的一些问题进行了探讨和改进：①为克服最小熵反褶积对弱反射的压制作用，本文采用对数型规范方差模准则；②为了减少最小熵反褶积对输入记录相位的改造，本文对反褶积算子加入了零相位约束；③为了克服局部极值问题并加速收敛，本文按输入记录傅氏谱的特征给出了确定迭代初始滤波器的方法。理论和实际资料处理表明，该方法是一种较理想、较实用的反褶积方法。 In this paper, some problems existing in the conventional method of least entropy deconvolution are discussed and improved: (1) In order to overcome the suppression effect of the minimum entropy deconvolution on the weak reflection, we adopt the logarithm-type norms variance model; (2) In order to reduce the minimum entropy In order to overcome the local extremum problem and speed up the convergence, this paper gives a method to determine the iterative initial filter according to the characteristics of input recorded Fourier spectrum Method of the Theoretical and practical data processing shows that the method is a more ideal and practical deconvolution method.