基于时间序列相似性匹配算法的地震预测研究
|
摘要
把匹配抽象时间序列相似性的方法引入到地震预报的应用中,结合大量地震历史源数据,地震领域的专家经验知识和相关成果基础上,提出了一种简化的抽象时间序列匹配模型。该模型在对海量数据进行预处理筛选的基础上再进行时间相似性匹配,增加了横向和纵向多方位地区和多方位时间段的匹配,不同时间差和阈值的匹配,并通过大量实验对该模型进行了反复验证,同时对我国地震频繁地区近几十年的地震历史数据进行了相似性匹配实验分析,取得了可信度较高的实验结果,实验结果验证了所给时间序列相似性匹配控制策略的有效性、实用性以及算法的优越性。
On the basis of analyzing the newly time sequence research achievement nowadays,several definitions on seismological zone relativity are put forward in this paper for integrating the large amount of history earthquake source data and the experimental expert knowledge in seismological field.At the same time,the time sequence similarity-matching model of the relevant seismological zone is presented,and then it is implemented through several correlative experimental simulations.Based on the sequence similarity-matching model,a sequence-matching algorithm is given with seismological similarity.Furthermore,by discovering the history earthquake database in recent several years,some experiments are provided to analyze longitudinal thick-granularity sequential similarity and thin-granularity sequential similarity.Finally,the experimental result has found its satisfactory way out by using the proposed algorithm to support earthquake prediction.
On the basis of analyzing the newly time sequence research achievement nowadays,several definitions on seismological zone relativity are put forward in this paper for integrating the large amount of history earthquake source data and the experimental expert knowledge in seismological field.At the same time,the time sequence similarity-matching model of the relevant seismological zone is presented,and then it is implemented through several correlative experimental simulations.Based on the sequence similarity-matching model,a sequence-matching algorithm is given with seismological similarity.Furthermore,by discovering the history earthquake database in recent several years,some experiments are provided to analyze longitudinal thick-granularity sequential similarity and thin-granularity sequential similarity.Finally,the experimental result has found its satisfactory way out by using the proposed algorithm to support earthquake prediction.
引文
[1]张保健,何华灿.时态数据挖掘研究进展[J].计算机科学,2002,29(2):124-126.
[2]欧阳为民,蔡庆生.数据库中的时态数据挖掘研究[J].计算机科学,1998,25(4):60-63.
[3]刘念祖.时态数据挖掘的探讨[J].上海第二工业大学学报,2001,(2):27-31.
[4]张保健.时间序列数据挖掘[D].西安:西北工业大学(博士学位论文),2003.
[5]欧阳为民,蔡庆生.在数据库中自动发现广义序贯模式[J].软件学报,1997,8(11):864-870.
[6]王炜,刘悦,李国正,等.中国大陆强震时间序列预测的支持向量机方法[J].地震,2005,(4):26-32.
[7]王炜,谢端,宋先月,等.使用人工神经网络进行我国大陆强震时间序列预测[J].西北地震学报,2002,24(4):315-319.
[8]崔万照,朱长纯,保文星,等.混沌时间序列的支持向量机预测[J].物理学报,2004,53(10):3303-3309.
[2]欧阳为民,蔡庆生.数据库中的时态数据挖掘研究[J].计算机科学,1998,25(4):60-63.
[3]刘念祖.时态数据挖掘的探讨[J].上海第二工业大学学报,2001,(2):27-31.
[4]张保健.时间序列数据挖掘[D].西安:西北工业大学(博士学位论文),2003.
[5]欧阳为民,蔡庆生.在数据库中自动发现广义序贯模式[J].软件学报,1997,8(11):864-870.
[6]王炜,刘悦,李国正,等.中国大陆强震时间序列预测的支持向量机方法[J].地震,2005,(4):26-32.
[7]王炜,谢端,宋先月,等.使用人工神经网络进行我国大陆强震时间序列预测[J].西北地震学报,2002,24(4):315-319.
[8]崔万照,朱长纯,保文星,等.混沌时间序列的支持向量机预测[J].物理学报,2004,53(10):3303-3309.
版权所有：© 2023 中国地质图书馆 中国地质调查局地学文献中心