摘要
将位场向下延拓作为一种反问题,结合正则化方法和快速傅里叶算法的优点,本文提出了一种解决位场向下延拓问题的迭代最小二乘法,新方法将位场向下延拓问题转化为最小二乘问题,采用梯度下降法对最小二乘问题进行迭代求解,在迭代过程中,利用快速傅里叶变换方法解决了计算量大的问题,新方法对延拓数据中的噪声具有很好的抑制作用,利用不合噪声和含有噪声的模型数据对新方法进行了验证,结果表明新方法快速、稳定、抗干扰能力强,向下延拓距离大.
To treat potential field downward continuation problem as an inverse problem and to make full use of advantage of regularization method and fast Fourier algorithm, this paper presents an iterative least square method to deal with downward continuation of potential field. The presented method converts downward continuation problem to a least square problem, and adopts gradient descent iteration method to search for the solution of the least square problem. In the process of iteration, fast Fourier algorithm is used to deal with computation problem. New method is not sensitive to the noise in original data. Model data with and without noise is simulated to test the presented method. The result shows that the new method is fast, stable, robust and has a large downward continuation distance.
出处
《地球物理学进展》
CSCD
北大核心
2011年第3期894-901,共8页
Progress in Geophysics
基金
国家自然科学基金项目(40774059)
教育部博士点基金项目(20070335090)联合资助
关键词
位场向下延拓
迭代最小二乘法
正则化方法
梯度下降法
快速傅里叶算法
potential field downward continuation, iterative least square method, regulation method, gradient descent method, fast Fourier algorithm
作者简介
陈龙伟，男，1982年生，河北藁城人，国防科技大学博士生，主要从事地磁数据处理、组合导航研究（E-mail：chenlongwei1982@sina．com）
