计算机与现代化

• 人工智能 • 上一篇    下一篇

基于频繁项目集的震前遥感异常信息挖掘

  

  1. (南京航空航天大学计算机科学与技术学院,江苏 南京 210016)
  • 收稿日期:2016-01-26 出版日期:2016-08-18 发布日期:2016-08-11
  • 作者简介:李文(1990-),男,安徽天长人,南京航空航天大学计算机科学与技术学院硕士研究生,研究方向:数据挖掘,建模与仿真; 徐慧(1989-),女,安徽池州人,硕士研究生,研究方向:数据挖掘,建模与仿真。

Pre-seismic Anomalies Detection with Frequent Itemsets from Remote Sensing Satellite Data

  1. (College of Computer Science and Technology, Nanjing University of Aeronautics and Astronautics, Nanjing 210016, China)
  • Received:2016-01-26 Online:2016-08-18 Published:2016-08-11

摘要: 已有研究表明地震活动区域红外遥感数据存在异常。本文针对遥感数据提出一种数据挖掘方法,运用频繁项集挖掘地震区域数据的异常特征,然后进行地震预测,同时选取非震数据进行反向验证。AIRS)2006~2013年的数据进行分析,根据大量实验,初步得出大气红外探测器数据可用于地震预测,地震预测最佳前兆时间为30天,前兆区域为边长2°正方形,预测的准确率最高可达81.8%,误报率为5.6%。使用大气红外探测器(

关键词: 数据挖掘, 地震预测, 红外遥感, 频繁项集

Abstract: Existing researches showed that seismic activity area remote sensing data was anomalous before an earthquake. In this paper, we proposed a method of data mining that could analyze the hyper-spectrum data from AIRS (atmospheric infrared detector) from January 2006 to December 2013. The main idea of this method was that we mined frequent itemsets of pre-seismic data and regarded them as pre-seismic anomalies, then we used the model to predict earthquakes. At the same time, we carried out reverse experiments using non-seismic data to verify this method. Main experimental results revealed that the AIRS data can be used to predict earthquakes. Among all experiments, the best prediction effect comes out when we set precursor time as 30 days and precursor region as square with center being epicenter and side length being 2°. The prediction accuracy is 81.8% and false positive is 5.6%.

Key words: data mining, earthquake prediction, infrared remote sensing, frequent itemsets

中图分类号: