摘要
主要介绍时序模式基本算法与原理，通过对地形变数据进行反正切矢量化，求出整时值（每小时记录一个数据）的反正切值。其后推算出每个月的反正切值，对其每个月的相似误差进行拟合与分析，得出偏离1．2倍中误差的月份。通过对大姚2次6级以上的地震对应性分析得出，震前1-5个月有明显的异常现象，以此说明该方法在地形变数据处理中具有很好的实用性。
The author mainly introduces the basic principles and algorithms of time-series model. By processing the crustal deformation data to arctangent vector, the hour data of the arctangent is calculated （one data per hour）, and then the arctangent data of each month is extrapolated. Through fitting and analyzing the similar error of every month, the author obtain the months failing away from the 1.2 times medium error threshold. The corresponding analysis of Dayao 6.0 double earthquakes shows an abnormal phenomenon in one to five months before the earthquakes, which illuminates that this method can be practical in the crustal deformation data processing.
出处
《内陆地震》
2010年第2期137-143,共7页
Inland Earthquake
基金
国家质检公益性行业科研专项(10-215)
关键词
数据挖掘
时序模式
地形变
Data mining
Time-series pattern
Crustal deformation
