运动目标地震动信号的时频特征分析

Eigenvector Analysis in Time and Frequency of Seismic Signals of Moving Target

该文对由典型地面车辆目标———轮式车、履带式车引起的地震动信号进行了实时探测 ,对实验所得的信号 ,应用短时傅立叶变换、小波及小波包分析方法对信号进行了处理 ,得到了时频分布矩阵奇异值分布特征 (SVD)和小波及小波包分解能量分布特征 (WWDD)。采用改进的BP网络 ,对远距离目标的地震动信号进行目标识别 ,应用WWDD对远距离信号的识别率可达 85%以上 ,说明WWDD具有更好的可分性。

A great number of seismic signals caused by moving wheeled vehicle and tracked vehide are obtained from field testing,and they have been processed applying Fourier transform,wavelet and wavelet package transform. On the basis of processing results, target recognition has been made by improved BP neural network, the result indicates that the recognition portion is as high as 85% on the eigenvector of wavelet and wavelet package decompose (WWPP),and shows that the eigenvector of WWPD has better performance.