基于小波包变换的地震事件分类
|
摘要
研究了地震信号在小波包变换下的特性,依据地震事件识别中"历史事例对比法"的思想,根据不同震源地震信号频率时变特性的不同,提出了基于"能量-事件"的地震事件分类方法。该方法不需要系统的模型结构,而是直接利用各频率成分能量的变化来进行事件识别,避免了对地震信号、传播途径准确建模的困难,简便、直观地完成了事件的识别。采用统计方法对实验样本进行处理,分别拟合出某地震台站记录的某地区天然地震、核爆炸事件的特征值和容差范围,并据此对检验信号集进行了检验,实验证明,该方法的事件识别率可达到95%以上。
The feature of seismic signal under wavelet packet transform is discussed in this paper. According to the thinking of “the method to contrast historical examples” in seismic event recognition, an “Power-Event” classifying approach is put forward, which can recognize seismic signals from difference hypocenter depending on the feature that their frequencies vary with time. Directly using the power changing of all frequency components, it can carry out the event recognition simply and conveniently without constructing accurate models for the seismic signal and the transmit approach. In the experiment, the statistical method is adopted to calculate the feature value and the permitted error range of the natural seismic signal and the nuclear explosion signal, by dealing with a great number of the specified area's data samples, which are recorded by specified site. The test simple set is recognized by above result. It is proved to be an effective seismic signal recognition approach, in which the event recognition rate can be reached 95%.
The feature of seismic signal under wavelet packet transform is discussed in this paper. According to the thinking of “the method to contrast historical examples” in seismic event recognition, an “Power-Event” classifying approach is put forward, which can recognize seismic signals from difference hypocenter depending on the feature that their frequencies vary with time. Directly using the power changing of all frequency components, it can carry out the event recognition simply and conveniently without constructing accurate models for the seismic signal and the transmit approach. In the experiment, the statistical method is adopted to calculate the feature value and the permitted error range of the natural seismic signal and the nuclear explosion signal, by dealing with a great number of the specified area's data samples, which are recorded by specified site. The test simple set is recognized by above result. It is proved to be an effective seismic signal recognition approach, in which the event recognition rate can be reached 95%.
引文
[1]MallatSG.Multifrequencychanneldecompositionsofimagesandwaveletmodels[J].IEEETrans.ASSP,1989,37(12):2091.
[2]DaubchiesI.Orthogonalbasesofcompactlysupportedwavelets[J].CommunicationPureandApplicationMathematics,1988,41:909.
[3]CoifmanRR.Adaptedmultiresolutionanalysis,computation,signalprocessingandoperatortheory[M].ProceedingoftheInternationalCongressofMathematicians,Kyoto,MathSocJapan,1990,887.
[4]林俊德,地震事件识别研究的若干建议.个人通信,1995.
[5]奥·克德罗夫.远震距离上短周期P波的地下核爆炸当量估算[R].苏联科学院报告,1988,300(3):
[6]范万春,等.远震P波群信号的自适应时-频分析[J].计算物理.2000,17(1-2):95.
[2]DaubchiesI.Orthogonalbasesofcompactlysupportedwavelets[J].CommunicationPureandApplicationMathematics,1988,41:909.
[3]CoifmanRR.Adaptedmultiresolutionanalysis,computation,signalprocessingandoperatortheory[M].ProceedingoftheInternationalCongressofMathematicians,Kyoto,MathSocJapan,1990,887.
[4]林俊德,地震事件识别研究的若干建议.个人通信,1995.
[5]奥·克德罗夫.远震距离上短周期P波的地下核爆炸当量估算[R].苏联科学院报告,1988,300(3):
[6]范万春,等.远震P波群信号的自适应时-频分析[J].计算物理.2000,17(1-2):95.
版权所有：© 2023 中国地质图书馆 中国地质调查局地学文献中心