基于STFT和CNN的齿轮箱故障诊断

Fault Diagnosis of Gearbox Based on STFT and CNN

针对浅层机器学习方法应用于齿轮箱故障诊断故障识别率低的问题,提出一种基于短时傅里叶变换和卷积神经网络的齿轮箱智能故障诊断方法。对齿轮的振动信号进行短时傅里叶变换得到时频图并输入到CNN故障诊断模型,根据模型输出的结果给出齿轮箱的故障状态,从而实现齿轮箱的故障诊断。在齿轮箱动力学模拟实验台采集多种不同故障齿轮的振动信号进行实验验证。实验结果表明:该方法能有效识别齿轮的故障状态,故障诊断准确率能够达到100%。

To solve the problem of low fault recognition rate of shallow machine learning methods applied to gearbox fault diagnosis,a gearbox intelligent fault diagnosis method based on short-time Fourier transform and convolutional neural network is proposed.Short-time Fourier transform is performed on the gear vibration signal to obtain the time-frequency diagram and input it into the CNN fault diagnosis model.According to the output of the model,the fault status of the gearbox is given,so as to realize the fault diagnosis of the gearbox.The vibration signals of a variety of different faulty gears are collected on the gearbox dynamics simulation experiment platform for experimental verification.The experiment results show that the method can effectively identify the fault state of the gear with the fault diagnosis accuracy rate at 100%.