论文部分内容阅读
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 issue in the interpretation of automatic seismic data is the initial phase identification using data recorded by a three-component station. In this paper, three-component stations of ARCESS, NORESS, FINESA and GERESS arrays as well as Polish KSP and former Soviet Union GRAM III Phase-polarization property of component station recorded data A 3-layer BP neural network is designed to recognize the seismic facies. Due to the multi-dimensionality of input data and its dependence on stations, this method solves the traditional Problems in the method.