发动机异响信号的小波包能量特征提取

Wavelet Packet Energy Feature Extraction of Engine Unusual Noise

在机械故障诊断过程中,最关键的问题就是故障特征信号的特征提取,从某种意义上说,特征提取是当前机械故障诊断研究中的"瓶颈"问题。发动机是一种多振源、宽频带、振动形态复杂的机械,其振动信号呈现非平稳时变特征,噪声干扰大,故障信号往往被淹没在干扰噪声中。发动机声响的分析在其故障诊断中显得极为重要。现提出一种依靠小波包分析来进行发动机故障诊断的方法,即通过对发动机异响信号在全频带范围内进行正交小波包分解,得到由全频带均匀划分的各子频带的小波包分解系数,对小波包分解系数进行重构得到该频带的信号,提取各频带信号的能量构造出小波包特征向量,从而实现对故障源的判断。

It is known that feature extraction is the most important and difficult topic in the field of mechanical fault diagnosis. To some extent, feature extraction is a choke point of the mechanical fault diagnosis technique. The engine is large machinery with multiengine vibration source, broadband and complex vibratile patterns,its vibratile signals show non-stationary,time-varying characteristics. The signal of failures is often drowned in the noise disturbance. So analysis, of engine noise is particularly important in the diagnosis of the engine failure. This paper presents engine fault diagnosis methods, according to a package of wavelet analysis. It is that the wavelet packet coefficient received from the entire band uniform of the various bands through the engine abnormal sound signal in the entire band within orthogonal wavelet packet. The Band signal can be reconstructed from the wavelet packet coefficient,and pick-up of the energy of various band signal can construct the wavelet packet feature vectors, so we can actualize the judgment from the source of the failure.