摘要
针对传统经验模态分解EMD时频分析功能不足的缺陷,提出了基于经验模态分解EMD和频率切片小波变换FSWT组合的爆破振动信号分析方法。对实际工程采集到的爆破振动信号进行EMD分解,根据相关性系数确定优势分量实现信号重构,并获取重构信号全频带FSWT时频特征。利用FSWT逆变换能切割任意频率区间的特点,将重构信号选择时间、频率切片区间进行了更为细化时频特征提取。研究了EMD-FSWT组合方法、Hilbert-Huang变换(HHT)、小波变换(WT)三种方法的消噪滤波效果,并与短时Fourier变换(STFT)、重排平滑Wigner-Ville分布(RSPWVD)两种传统时频方法进行了对比。分析结果表明:EMD-FSWT组合方法,对瞬态信号在时频域上的分辨率更高,消噪和滤波效果好,适于对爆破振动信号进行更为精细化的时频特征分析。
In view of the shortcomings of the traditional empirical time-frequency analysis methods, a blasting vibration signal analysis method based on EMD-FSWT was proposed. The EMD of blasting vibration signals collected in actual engineering was carried out, according to the correlation coefficient, to determine the dominant component and reconstruct the signal, and the full band FSWT time-frequency characteristics of the reconstructed signal were then provided. In virtue of the characteristic of the inverse transformation of FSWT, that any frequency range of the signal can be cut out, the time and frequency ranges of the reconstructed signal were chosen and more refined time-frequency feature extraction was achieved. The filtering de-noising effects of the EMD-FSWT combination method, Hilbert Huang transform (HHT) and wavelet transform (WT), were analysed and compared with those of two traditional methods, the short time Fourier transform (STFT) and rearrangement smooth Wigner-Ville distribution (RSPWVD). The analysis results show that: the EMD-FSWT combination method has higher time-frequency domain resolution in transient signal analysis. Its de-noising and filtering effect is good and is more suitable for precise time-frequency characteristics analysis for the blasting vibration signal. © 2017, Editorial Office of Journal of Vibration and Shock. All right reserved.
出处
《振动与冲击》
EI
CSCD
北大核心
2017年第2期58-64,共7页
Journal of Vibration and Shock
基金
国家自然科学基金-面上项目(51274203)
关键词
爆破振动
经验模态分解
频率切片小波
时频分析
能量分布
Blasting
Feature extraction
Frequency domain analysis
Inverse problems
Mathematical transformations
Signal processing
Vibration analysis
Wavelet analysis
Wavelet decomposition
Wavelet transforms
Wigner
Ville distribution
