Research Of Regional Correlation Based On Natural Seismic Data

So far,in the world of scientific research,earthquake prediction is still a problem.China is located between the Pacific Rim seismic belt and the Eurasian seismic belt,by the Pacific plate,the Indian plate and the Philippine sea plate extrusion,seismic fault zone is very active,belonging to the earthquake frequent countries.Seismic activity has a large impact,and the earthquake will lead to aftershocks and secondary disasters,which pose a serious threat to human life and property security.Therefore,the study of earthquakes has great value.Seismic data is different from the previous data,often complex structure,changing form,with a high degree of non-linear characteristics,and with the advent of the information age,the rapid increase in the amount of seismic data,the source of the earthquake more diversified.The traditional means of data analysis are time-consuming and laborious,have been unable to meet the demand of seismic depth data mining.Many research results have shown that the seismic activities between adjacent areas and seismic zones,and even ultra-long distance seismic zones,have a certain correlation in time and space.The purpose of this thesis is to apply the temporal data mining method to the study of natural seismic data,so as to find the relationship between the seismic areas and the inter-regional timing relationship,and then provide the relevant theoretical basis for the earthquake prediction.The traditional time series similarity measure methods can achieve good results for dense and continuous numerical time series,but the seismic event are small probability events.The seismic data set is often sparse in a certain time range.Because of this sparse characteristic,So do not have rich frequency domain characteristics and trends,shape characteristics.Based on the existing time series similarity measure methods and the characteristics of seismic data,this thesis redefines the time series similarity measure model and the time series similarity matching algorithm.In order to carry out the data mining smoothly,this thesis firstly analyzes the problems existing in the seismic data,and uses the data reduction and data cleaning technology to achieve the attribute reduction,missing value filling,duplicate value removal,and geographically Location information added to complete the data preprocessing work.Then,the seismic data set is abstracted as a set of block seismic time series,and the aftershocks events are removed.Finally,the time series similarity measure model defined in this thesis is applied to a large number of historical seismic data,and the time series matching algorithm is realized by experiments with different spatial scale,time window scale and magnitude scale.The experimental results show that the proposed algorithm can effectively find the region with high correlation of seismic activity,and there are obvious characteristics of unidirectional sequence relationship between the regions with high correlation,which is of great significance to the earthquake prediction research.