A Research On Algorithms For Seismic Wave Feature Extraction By The Time Window Method

Seismic(vibration)waves refer to the vibrational energy transmitted in the earth's solid medium and originate from various natural or artificial vibration source events.Seismic waves originate from natural seismic events or artificial blasting events and carry internal information such as sources and wave propagation paths.It is a reliable carrier for human understanding of the complex structures inside the Earth.As humans continue to increase their intensive activities on the Earth,the range of activities and the intensity of their activities are also constantly expanding.The underground vibration caused by various human factors will also leave a mark in the detection and recording of seismic waveforms to some extent.The seismic signals recorded by the Seismological Observatory Network contain many non-natural earthquakes such as blasting in artificial mines and chemical explosions.The recording of these events,if not promptly removed,will disrupt the relevant seismic records and adversely affect the further development of seismology.On the contrary,if we can identify natural earthquakes and artificial blasting events from the waveforms detected at the seismic stations,we can accurately create natural earthquake events and catalogues of artificial blasting events to provide more reliable seismic research.data observation.This paper first introduces the experimental data and sources used,then preprocesses the extracted waveforms,and proposes a wavelet threshold denoising method based on EMD.Then from the perspective of scientific data visualization,we introduce the programming tools and specific implementation steps used during the experiment.Under the visualization study,this paper can present complex data information in a graphic image to highlight the information that one wishes to express.This has laid a foundation for the following experimental research.A new algorithm for automatic detection of different time windows is proposed: VAR-AIC algorithm.The specific windowing procedure is divided into two steps: First,set the seismic signal to the sampling rate of the same window length,and combine the VAR-AIC algorithm to detect the arrival of the P-wave and S-wave,and the waveform end time;then,set different windows.The number of superposition,the current seismic signal is divided into a specified window width according to the overlap window and specifies the sliding step window,thesliding step length is equal to the window length,and the sliding step length is equal to 4/5 times the length of the window are analyzed,respectively,to obtain the window length The total number of windows of seismic signals of step,step length and segmentation length,and then statistics of P wave arrival time and S wave arrival time,as well as waveform end time,can extract different wave characteristics.After a large number of experiments,the P wave is obtained.Identification criteria such as the number and amplitude of S waves.The experimental results show that compared with the existing methods of STA-LTA and AR-AIC,this method has a shorter computation time and can recognize the arrival of P-wave first arrival time earlier.Thus,the different features of natural earthquakes and artificial blasting are well represented.This method has a higher recognition rate.It is more suitable for application in seismic identification and provides a longer window period for earthquake warning work,thereby reducing the loss of people and property.Finally,in order to better improve the recognition rate and ensure stability,a Bayesian classifier and TKEO algorithm are combined to form a complex classifier with superior performance.The waveform features are identified successively,and a certain time window length is then selected for TKEO algorithm calculation.Then the TKEO algorithm values of different time window lengths are averaged,and the data normalization processing is performed.Finally,the number of P waves and S waves and the end of the waveform are significantly different,the correct recognition rate is high.