摘要
由S变换推导出的时序分解算法可以将一个任意的初始时间序列变换成一组突出时间序列局部信息的二维时间序列,该时序分解的可逆性表明了它可用于时域信号的滤波与特征提取。希尔伯特变换可有效地对时域信号进行解调,其实质是对原始信号作一次特殊的滤波。综合前述两种变换的优点,提出了结合希尔伯特变换及时序分解的弱故障特征信号提取算法,采用数值仿真实验及齿轮故障诊断进行了验证,结果表明,此种方法能有效地提取混在强背景信号中的弱故障特征信号。
The time series decomposition algorithm that is deduced by S transform can be employed to decompose an arbitrary initial time series into a two-dimensional time series, through which much prominence is given to the local message of the original time series. The convertibility of this decomposition indicates that it can be used for the time domain signal filtering and feature extraction. Hilbert transformation can be use to demodulate the time domain signal effectively, which executes a special filtering to original signal essentially. Combining the advantages of those two kinds of transforms, an algorithm for extracting the weak fault feature signal is proposed. The result received by numeric value simulation and the gear fault diagnosis experiment indicates that this algorithm is able to extract the weak fault feature signal mixed in the powerful backdrop signal effectively.
出处
《振动工程学报》
EI
CSCD
北大核心
2007年第1期24-28,共5页
Journal of Vibration Engineering
基金
湖北省自然科学基金资助(2005ABA287)
关键词
故障诊断
弱信号
齿轮
包络分析
时间序列分析
fault diagnosis
weak signal
gear
enveloped analysis
time series analysis
