Abstract:In this paper, the authors used the waveform records of the 2006 to 2017 earthquake events in Shandong province, and the wavelet transform of three kinds of earthquake type: natural seismic, blasting and collapse waveform are carried out, and Shannon entropy features are extracted by support vector classifier LIBSVM. A series of experiments are designed to study the factors that affect the final classification effect. The results show that the length of signal window, the way of wavelet decomposition, the type of wavelet base, the type of vector machine and the type of vector machine kernel function all have some influence on the result of seismic classification. The combination of 2 000 seconds signal window length +db7 wavelet base+υ-SVC vector machine is used in the three groups with the highest recognition rate. The combination of several factors with high recognition rate can be applied to real-time recognition of earthquake type in the future to further improve the recognition rate of earthquake type and trigger accuracy.
FAN Xiaoyi,QU Junhao,LIU Fangbin et al. Analysis of Influencing Factors in Use of Support Vector Machine Method to Identify Earthquake Types[J]. jgg, 2020, 40(10): 1034-1038.