震源识别中小波变换类型的选择研究

Choice of wavelet transform type for automatically recognition of earthquake and explosion

在天然地震与人工爆破的识别中,小波变换凭借它良好的表征信号时域和频域局部特征的能力得到了人们的青睐。探讨了离散小波变换、静态小波变换和小波包变换在天然地震与人工爆破识别中的适用性问题,分类效果的检验采用的是-νSVC算法。实验结果证明,只要选择合适的小波基函数,从它们变换后的小波系数中提取出来的香农熵特征都能很好地表达天然地震与人工爆破之间的本质区别。而且从实验结果中还可以看出,随着特征向量维数的增加,分类准确率有提高的趋势,但是整个识别过程所花费的时间也将随之增加。

In the recognition of earthquake and explosion,wavelet transform receives widely attentions due to its capability of simultaneous depicting wave signal characteristics in time and frequency domain.This paper discusses on the applicability of DWT,static wavelet transform and wavelet packet in the automatically recognition of earthquake and explosion,the classification results of the tests used ν-SVC algorithm.The results showed that if appropriate wavelet base function being used,any of these three types of wavelet transform could be applied to discriminate earthquake and explosion by Shannon entropy feature established from wavelet coefficients.Furthermore,the experiments results also reveal that the higher dimension of the feature vector,the higher correct recognition rate,although the recognition processing time also obviously increases.