地震动信号的分析与目标识别
Seismic Signals analysis And Identification
-
摘要: 通过外场实验获得关于轮式车、履带式车的大量地震动信号,在时-频域应用多种方法对信号进行处理,得到相应的特征矢量。利用改进的BP网络对远距离的地震动信号进行目标识别,基于小波及小波包分解能量分布特征的识别率可达85%以上,这种特征矢量具有较好的可分性。Abstract: Great deals of signals obtained from field-testing have been processed in time and frequency domain with Short Time Fourier Transform and wavelet package transform techniques. On the basis of far distance seismic signal, target recognition has been made by BP neural network, the result indicates that the recognition ratio is as high as 85% on the eigenvector of wavelet package decompose, and shows that the eigenvector of wavelet package decompose has better performance.