基于一维卷积神经网络的多工况轴承故障诊断

Bearing fault diagnosis in multiple working conditions based on one dimensional convolutional neural network

针对多工况下的轴承故障诊断问题,提出一种基于一维卷积神经网络的故障诊断方法。采用重叠采样方法从CWRU数据集获得扩充的数据样本,并通过归一化获得训练数据;基于TensorFlow模型设计了轴承故障诊断的一维卷积神经网络,将预处理后的数据样本直接输入到网络中进行训练,实现了端到端的轴承故障诊断。实验表明,该方法避免了复杂的特征提取过程,具有较高的故障分类准确率和鲁棒的泛化能力,可以实现多工况下轴承故障的准确诊断。

To deal with the bearing fault diagnosis in multiple working conditions,this paper proposed a fault diagnosis method based on one-dimensional convolutional neural network.The overlapped sampling method was first used to obtain the augmented data samples from the CWRU data set,which were then normalized to obtain the training data.Based on the TensorFlow model,the one-dimensional convolutional neural network was designed for the bearing fault diagnosis,into which the preprocessed data samples were input directly for training,thus accomplishing the bearing fault diagnosis in an end-to-end manner.The experiments show that the proposed method,which can avoid the complex feature extraction procedure and achieve higher fault classification accuracy and robust generalization capability,can achieve the accurate bearing fault diagnosis in multiple working conditions.