论文部分内容阅读
本文介绍了地震波的句法模式识别方法(也称结构模式识别方法),建立了一个用该技术分析地震波的完备系统。地震波可以分割成相同长度的若干段,然后转换成一串基元组成的链。地震波识别系统可以由模式表示子系统、自动文法推断子系统和一个最小代价误差校正剖析器组成;也可以用最近邻判决规则对模式表示子系统的输出进行分类而直接构成。在具体细节上,基元的选择采用聚类方法;最近邻判决过程中,链之间距离的计算采用了动态规划技术。文章简略讨论了各个环节的一些基本问题,给出了天然地震和人工爆炸资料的试验结果。
This paper introduces the method of syntactic pattern recognition of seismic waves (also called structural pattern recognition method), and establishes a complete system for seismic wave analysis using this technique. Seismic waves can be divided into several segments of the same length and then converted into a chain of elementary elements. The seismic wave recognition system can be composed of a pattern representation subsystem, an auto-grammar deduction subsystem and a minimum cost error correction profiler. The nearest neighbor decision rule can also be used to classify the output of the pattern representation subsystem directly. In specific details, the choice of primitives using clustering method; nearest neighbor decision process, the calculation of the distance between chains using dynamic programming techniques. The paper briefly discusses some basic problems in each link and gives the test results of natural earthquake and artificial explosion data.