论文部分内容阅读
电机滚动轴承的时域振动信号经过快速傅里叶变换和自功率谱处理后，可以获得电机滚动轴承振动的固有频率，然后运用该频率指标，利用多层反向传播前馈型神经网络，通过神经网络的学习和推广两个阶段，可以实现对轴承故障的自动分类诊断。对一电机滚动轴承的实验表明，该方法行之有效，对工程应用具有较高的实用价值。与其它损伤识别指标相比，诊断精度相对提高率均大于１１％，说明频率指标对结构的损伤具有更高的灵敏度。
After the vibration signal of the motor rolling bearing is processed by fast Fourier transform and self-power spectrum, the natural frequency of vibration of the motor rolling bearing can be obtained. Then by using the frequency index, using multi-layer back propagation feedforward neural network and neural network Learning and promotion of two stages, you can achieve the automatic classification of bearing fault diagnosis. Experiments on a motor rolling bearing show that the method is effective and has high practical value for engineering application. Compared with other damage identification indexes, the relative improvement rate of diagnosis accuracy is more than 11%, indicating that the frequency index has higher sensitivity to structural damage.