基于卷积神经网络的高速列车抗蛇行减振器故障诊断

由于抗蛇行减振器在服役过程中发生故障会严重威胁到列车的运行安全，提出一种基于卷积神经网络的抗蛇行减振器故障诊断方法。采集的阻尼力信号通过短时傅里叶变换得到时频图谱，并将其划分为训练集和测试集；然后将训练集输入卷积神经网络模型中，进行训练样本信号的特征提取工作，通过前向传播和反向传播方式得到卷积神经网络的具体模型，并通过多次迭代更新网络参数；最后，将训练好的模型用于测试集，获得蛇行减振器故障诊断的准确率。为了验证模型的有效性，选用正常状态、启动不良和锯齿波故障数据作为实验验证，结果表明：该方法不仅避免了传统诊断方法需要人工提取特征带来的问题，且具有较好的诊断效果。

Fault Diagnosis of Yaw Damper in High-Speed Train Based on Convolutional Neural Network

Since the fault of the yaw damper during service will seriously threaten the safety of train operation,a fault diagnosis method for the yaw damper was proposed based on convolutional neural network. The time-frequency spectrum of the collected damping force signal was obtained by short-time Fourier transform, and it was divided into a training set and a test set. Then, the training set was input into the convolutional neural network model, and the characteristics of the training sample signal were carried out. Further, the specific model of the convolutional neural network was obtained through forward propagation and back propagation, and the network parameters were updated through multiple iterations. Finally, the trained model was used for the test set to obtain the fault diagnosis accuracy of the yaw damper. In order to verify the validity of the model, the data of normal state, poor startup and sawtooth wave fault were selected as experimental verification. The results show that the proposed method not only avoids the problems caused by the need for manual feature extraction, but also has a better diagnosis effect.