摘要
把神经网络与重磁异常反演理论相结合，提出了用于重磁反演的一种拟BP神经网络方法．基于3层神经网络结构，把隐含层神经元设定为三维空间物性（磁化强度或密度）单元．对实测与理论重磁异常经S型函数变换，采用自动修改物性单元物性值的拟BP算法，反演三维空间的物性分布．利用该网络对理论模型数据和内蒙古某花岗岩体上的航磁资料进行了反演计算，取得了满意的反演效果．
We combine neural network theory with inverded theory of gravity and magneticanomalies and Present the pseduo-BP netal netWork (PBP) method applied toinversion for gravity and magnetic anomalies. Based on the structure of fore layernetal network, the hidden layer neuron are hypothesised as physical property units(magnetization or density) in the 3D cuse. For measured and theoretical gravity andmagnetic anomalies through the translation of S--type function, the PBP algorithmupdating automaticly the values of physical property units can be applied to theinversion of the 3D spatial distribution of physical property. Applicahon of the PBPmethod to the inversion of synthetic data and aeromagnetic data from some gdriticbody in inner Mongolia has yielded the reasonable results.
出处
《地球物理学报》
SCIE
EI
CAS
CSCD
北大核心
1998年第2期242-251,共10页
Chinese Journal of Geophysics
基金
国家自然科学基金!49174205
地质矿产部科技司资助
关键词
重磁反演
拟BP神经网络
模型计算
磁法勘探
Gravity and magnetic inversion, PBP neural network, Modelcomputation, Applied case.
