摘要
发动机的振动信号一般呈非平稳时变特征。傅立叶变换只能确定一个函数奇异性的整体性质,而难以确定奇异点在空间的位置及分布情况。经典的小波变换具有空间局部化性质,但多数性质都是通过傅立叶变换性质来实现的。提升算法摆脱了这种局限,它从一个简单的多分辨分析开始,然后向具有某一特性的MRA逐渐逼近。根据信号与噪声在各尺度下的不同传播特性,通过各层之间相关性分析滤除噪声,有效的提取出信号特征。
The engine vibration signal usually shows non-stationary time varying characteristic. Fourier transform only makes sure the feature of a function singular and can not confirm the position and of singular point on space. Traditional wavelet transform has space local character. But most of feature wavelet are based on Fourier transform. Lifting scheme gets rid of the localization. It begins from a simply MRA, and then approach to a MRA with certain character gradually. According to the different spread feature between signal and noise on each scale, we make use of correlation analysis to removal noise and extract signal feature effectively.
出处
《噪声与振动控制》
CSCD
北大核心
2006年第3期32-34,共3页
Noise and Vibration Control
关键词
振动与波
提升小波
发动机
相关性
去噪
vibration and wave
lifting scheme
engine
correlation analysis
removal noise
作者简介
王锋（1980-）,男,山西太原人,硕士研究生,主要研究方向：信号处理及故障诊断理论研究.
