论文部分内容阅读
引入图像分析方法,提出了直接从转子故障信号连续小波尺度谱中提取图像纹理特征的新方法.首先,通过转子故障模拟实验台采集了不平衡、不对中、碰摩及油膜涡动等典型故障信号;然后,分析了故障信号尺度谱的差别及所提取出的数字特征对故障的敏感性;最后用结构自适应集成神经网络进行了智能诊断实验,结果表明了本文所提出的尺度谱数字特征对转子故障诊断的有效性.
A new method of image texture feature extraction from continuous wavelet scale spectrum of rotor fault signal is proposed by introducing the image analysis method.Firstly, typical faults such as unbalance, misalignment, rubbing and oil film whirling are collected by rotor fault simulation bench Then, the difference of the scale of the fault signal and the sensitivity of the extracted digital features to the fault are analyzed. Finally, an intelligent diagnosis experiment is carried out by using the structure-adaptive integrated neural network. The results show that the digital characteristics of the scale spectrum presented in this paper Effectiveness of rotor fault diagnosis.