| In the process of the continuous evolution of the Earth,the occurrence of earthquake is unavoidable.Seismic activity is often potentially accompanied by strong destructiveness.If earthquakes occur in densely populated areas,there will be huge loss of human life and property.The only way to monitor seismic activity is through vibration waveform signals.Humans have already taken measures to monitor seismic activities,such as establishing an extensive network of seismic observation stations.Many non-natural earthquake events often occur on the Earth.For example,nuclear tests,chemical explosions,blasting,mine collapse,ground collapse,and even the collapse of some large-scale engineering projects caused by human activities can generate strong seismic waves and spread out to spread energy.With the gradual expansion of the coverage of the observatory network and the continuous improvement of the sensitivity of observation instruments,seismic monitoring instruments can easily detect the seismic waveforms of all these natural and non-natural seismic events.When an earthquake occurs,three elements of the earthquake can be obtained based on the observed waveform data of the seismic network: source,magnitude,and time of origin.The information on the internal structure of the Earth can only be obtained indirectly based on the relevant characteristics of the seismic wave inversion waveform propagation path.The focus of this study is the vibration waveform corresponding to the active event source,which needs to be automatically intercepted from the continuous signal observed in the station,and then based on the intercepted vibration waveform signal,study its waveform characteristics to distinguish the source type(this article focuses on distinguishing Natural earthquakes and explosions).In this paper,three kinds of single-channel seismic waveform fractal dimensions(R/S analysis(rescaled range analysis),spectral method and box-dimensional method,etc.)were compared,simulated three-dimensional dynamic waveform diagram,three-dimensional waveform fractal dimension and Problems such as fractal features and complexity sample identification are discussed.Aiming at the problem of seismic source-type recognition,this thesis attempts to carry out 3D seismic wave visualization research on 3 components(vertical,horizontal east-west,horizontal north-south)of single earthquake observatory station.The 3 components of the original observatory waveform in the whole process of an event are filtered separately.Then noises in the 3 components are filtered out and then plotted out in the form of 3D time series(filtered wave signals).The 3D waveforms drawn can be adjusted interactively with the distance,point of view and azimuth angle of the wave coordinate system.A fractal feature extraction algorithm for 3D waveform is developed.Using the best-known box dimension method in the 1D waveform fractional dimension calculation,the non-empty box number of the covering waveform is obtained on the 3D waveform,and the best scale-free region is obtained through the dual-logarithmic curve.Finally,the values of 3D waveform fractal are calculated for all observatory station of an event from the noise-free filtered seismic waves corresponding the whole process of the corresponding events.The fractal value of the 3D waveform of an event is represented by the mean value of the fractal values of the 3D waveform of all stations of the event.In order to intuitively express the difference between different source types(earthquake or explosion),the waveform complexity of the vertical component in each observatory station is also calculated.The wave complexity of an event is represented by the average value of the waveform complexity of the vertical components of all the observatory stations of the event.The SVM method is used to classify and recognize the data sets of 35 natural earthquake events and 27 man-made explosion events.When the value of C σ equal to(1,0.01),the recognition rate of the dataset is the highest,up to 98.4%,and the efficiency of the algorithm is high.The experimental results show the validity of the 3D fractal feature proposed in this thesis. |
