摘要
地震勘探资料的噪声许多呈现混沌现象,利用传统的去噪方法效果并不理想.如何根据混沌固有的性质,对地震勘探资料中的有效信号进行提取是许多科学工作者极为关注的问题.针对这种混沌噪声下的微弱信号检测,本文提出三种神经网络方法并对此进行比较.理论分析及仿真实验表明这三种神经网络在信噪比达到-37dB时,均能检测混沌噪声背景中的微弱信号.
Most of noise mentioned in the data of seismic prospecting presents chaos, and it can not been removed ideally by traditional methods. It is a hot problem considered by many scientists that how to extract the valid signal in data of seismic prospecting with inherence of the chaos. To the detection of weak signal and the background of the chaos noise, there are three neural network methods are presented and compared in this paper. Theory analysis and simulation experiments show that these three neural network methods can all detect the weak signal embedded in the chaos noise background when SNR gets to -37dB.
出处
《地球物理学进展》
CSCD
2003年第4期743-747,共5页
Progress in Geophysics
基金
国家自然科学基金项目(40374045)资助.
