【摘要】 常见的机械故障诊断研究侧重于对故障的识别和分类 ,相应的故障诊断方法均为提高诊断的准确率而设计 ;从实际应用角度来讲 ,这样的诊断方法是不全面的。全面反映设备故障状况的因素除了故障类别外 ,还应指出故障的具体位置和程度。冲击、油膜振荡、碰摩和转速突变等故障往往产生奇异信号 ,奇异点包含了更为丰富的故障信息。小波分析具有良好的时频局部化特性 ,为描述信号的奇异性提供了手段。为此提出用小波分析方法 ,通过对奇异故障信号的检测、信噪分离和信号频带分析来提取故障特征 ,以确定故障的位置和程度。这种方法提取的故障信息应用在神经网络等其他故障诊断方法中可以更准确、更全面地诊断故障 ,柴油机和风机故障实例证明了该方法的有效性更多还原

【Abstract】 The research of diagnosing common mechanical fault usually put more emphasize on their recognition and classification. Thus the method of fault diagnosis is designed to increase diagnosis rate, while according to practical application, the method is not perfect. It is put forward, in this paper, that the reflection factors of mechanical failure are not only fault category, but also its position and intensity. Shock, oil whip, friction and rotating speed mutation produce singular signal, the singular point of signal contains enormous fault information. The wavelet analysis is a good way to describe singular signal, for it possesses a fine local time frequency characteristics. The feature of fault signal, which is obtained by singular signal inspection, signal noise separation and signal frequency range analysis by means of wavelet analysis, can be used to identify the position and intensity of the fault. The extracted fault information can be applied with other fault diagnosis methods, such as neural networks and fuzzy diagnosis. Two examples of diesel’s fault and blower’s fault show that the feature extraction method based on wavelet analysis is effective. This method strengthens the theory of fault diagnosis.更多还原