摘要
在自动地震数据解释中的一个重要问题是用三分量台站记录的数据来进行初始震相识别,本文利用ARCESS、NORESS、FINESA、GERESS台阵的三分量台站以及波兰KSP和前苏联GRAM三分量台站记录数据的震相偏振属性设计了一个3层BP神经网络,实现了对震相的识别,由于输入数据的多维度和对台站的依赖特性,该方法在一定程度上解决了传统方法中存在的问题。
An important problem in automatic seismic data interpretation is initial phase identification from data recorded by 3-component stations. This paper describes a 3-layer BP neural network approach to this problem, which uses the polarization attributes of data recorded by 3-component elements of the array stations ARCESS, NORESS, FINESA and GERESS, and 3-component stations in Poland (KSP) and in the former Soviet Union (GARM). Because of the high-dimensionality of the input, and station-dependent data, this approach solves several problems existed in traditional methods.
引文
[1] Dysart P, Pulli J. Regional Seismic Event Classification at the NORESS Array: Seismological Measurements and the Use of Trained Neural Networks. Bulletin of the Seismological Society of America, 1990, 80(6) : 1910-1933
[2] 刘希强,周蕙兰,等.基于小波包变换的弱震相识别方法.地震学报,1998,20(4) :373-380
[3] 胡守仁.神经网络导论.长沙:国防科技大学出版社,1993,113-120
[4] Jurkevics A. Polarization Analysis of 3-Component Array Data. Bulletin of the Seismological Society of America, 1988, 78 (5) : 1725-1743
[5] 蔡自兴,徐光佑.人工智能及其应用.北京:清华大学出版社,1996,206-217