基于深度信念网络的列车故障音频诊断方法

铁路在交通运输行业有着举足轻重的地位,一旦列车发生故障将会导致严重的生命财产损失。由于列车发生故障的概率相对较低,因此难以捕获列车的故障样本。针对上述问题,提出了一种无监督学习的列车故障识别方法,通过检测列车音频信号来识别列车故障。该方法基于深度信念网络(DBN),利用小波包分解提取检测信号的特征向量并将其作为DBN的输入,待网络充分训练后,由训练好的DBN识别当前列车的运行状况。现场监测实验结果表明,该方法能够在无监督的条件下有效识别列车故障,保障了列车的运行安全。

Audio Diagnosis Method for Train Fault Based on Deep Belief Network

Railway plays a leading role in the transportation industry.Once a train breaks down,it will cause serious loss of life and property.It is not easy to obtain train fault samples because of the low fault probability.In order to solve above problems,a train fault identification method without supervision is proposed,which can identify train faults by detecting train audio signals.Based on deep belief network (DBN),this method uses wavelet packet decomposition to extract the eigenvector of the detected signal,which is taken as the input of DBN to train the network.The trained network identifies the current train operation status.The field experimental results show that this method can effectively identify train faults and ensure the safety of the train without supervision.