Wavelet based compression and feature selection for vibration analysis
Wavelet based compression and feature selection for vibration analysis
复制标题
DOI：
10.1006/jsvi.1997.1380
复制
发表时间：
1998-04-16
影响因子：
4.7
通讯作者：
Staszewski, WJ
中科院分区：
文献类型：
--
作者：
Staszewski, WJ
研究了基于小波的线性变换在振动分析中的数据压缩和特征选择问题。综述了小波数据压缩的最新进展。讨论了各种类型的数据，包括周期性、连续非平稳和瞬时非平稳信号，用来展示小波压缩的实用方面。分析采用了光滑小波和紧支撑小波。结果表明，振动分析中的压缩不仅可以用于数据的有效存储和传输，还可以用于特征选择。已经提出了许多不同的方法来显示系数选择程序。这包括根据其幅度、位置和频率位置截断小波系数的程序，以及基于最佳小波系数的数据压缩技术。(C)1998年学术出版社有限公司。
This paper is concerned with wavelet based linear transformations for data compression and feature selection in vibration analysis. Recent developments in wavelet data compression are summarized. A discussion of various types of data including periodic, continuous non-stationary and transient non-stationary signals, are used to show practical aspects of wavelet compression. The analysis employs smooth wavelets and compactly supported wavelets. It has been shown that compression in vibration analysis can be used not only for effective storage and transmission of the data but also for feature selection. A number of different approaches have been presented to show coefficient selection procedures. This includes procedures based on truncated wavelet coefficients according to their amplitude, position and frequency location and a data compression technique based on optimal wavelet coefficients. (C) 1998 Academic Press Limited.