摘要
针对柴油机曲轴轴承磨损故障信号特征微弱,易被噪声湮没且不同故障程度信号较难区分的特点,提出了一种基于压缩小波和局部保持投影的柴油机信息熵增强方法。利用压缩小波对信号多尺度重构减弱噪声干扰,通过局部保持映射对多尺度信号进行降维,消除冗余信息并增强信号的冲击特性,最终以时域、频域以及时频域的三种信息熵表征信号特征。仿真和实例信号表明,该方法对故障信号特征增强明显,依据信息熵值实现了曲轴磨损状态的分类识别。
Aiming at that diesel engine crankshaft bearing wear fault signals' characteristics are very weak,they are easy to be buried by noise and signals of different fault levels are hard to be distinguished,a new information entropy enhancement method based on the synchro-squeezed wavelet transform and locally preserving projection was proposed.Firstly,a signal was reconstructed in multi-scale mode with the synchro-squeezed wavelet transform. Then the dimensions of the multi-scale signal were reduced with the locally preserving projection to eliminate its redundant information and enhance its impact characteristic. Finally,three information entropies in time domain,frequency domain and timefrequency domain were used to characterize the signal features. Simulated and actual signals showed that the proposed method can enhance the fault signals' features obviously,and realize classification recognition of crankshaft wear states according to information entropy values.
出处
《振动与冲击》
EI
CSCD
北大核心
2018年第3期180-185,共6页
Journal of Vibration and Shock
基金
后勤保障部重点项目(BS311C011)
关键词
局部保持映射
压缩小波
信息熵
特征增强
locally preserving projection
synchro-squeezed wavelet transform
information entropy
feature enhancement
