基于小波分析的汽轮发电机组振动信号消噪和特征提取

De-noising and feature extraction of vibration signals for turbogenerator units by using wavelet analysis

小波分析技术由于其良好的时频局部化性质,对突变和非平稳信号的分析具有良好的效果,已经成为信号消噪、特征提取和故障诊断的重要方法之一。针对汽轮发电机组的振动特征,采用基于最优小波包基的方法对汽轮发电机组的振动信号进行消噪处理,有效地剔除了汽轮发电机组表面振动信号的噪声干扰,提高了信号的信噪比;对消噪后的信号进行小波包分解,并将各相关频带进行能量特征提取,从而为汽轮发电机组振动信号的故障诊断提供了有力依据。

With good time-frequency localization and good analysis results of sudden-change signals and non-stationary signals, the wavelet analysis technology is one the important methods for signal de-noising, feature extraction, and fault diagnosis. Based on the vibration characteristic of turbo-generator units, the optimal wavelet package basis was used to de-noise the vibration signal of turbo-generator units, and the noise disturbance of the vibration signal is effectively eliminated, and the signal-noise ratio of the signal is improved. The de-noised signals were decomposed by wavelet package, and the energy feature was extracted from relevant frequency bands, which provides support for fault diagnosis of vibration signals for turbo-generator units.