卷积神经网络的神华关键配件状态自动跟踪研究

Research on Automatic State Tracking of Shenhua Key Parts Based on Convolutional Neural Network

为避免关键配件异常状态带来的神华铁路货车运行安全隐患,提出基于卷积神经网络的神华关键配件状态自动跟踪方法。神华关键配件状态图像作为卷积神经网络的输入数据,经过卷积和池化操作后获得神华关键配件状态检测结果,并将其作为图像第一帧的状态,然后利用核相关滤波训练获得的回归模型估计图像下一帧的状态,实现神华关键配件状态自动跟踪。实验结果表明:该方法能够获得较为完整、清晰的神华关键配件状态图像;不同神华关键配件状态检测的MCC值均在0.8以上,且能够在异常状态发生之前得到状态检测结果;各时刻的神华关键配件状态跟踪结果与实际结果完全相同。

In order to avoid the hidden danger of Shenhua railway freight train operation caused by the abnormal state of key parts,the automatic state tracking method of Shenhua key parts based on convolutional neural network is studied.The CCD camera is used to collect the state image of Shenhua key accessories,which is used as the input data of convolution neural network.After convolution and pooling,the state detection result of Shenhua key accessories is obtained,and it is used as the state of the first frame of the image.On this basis,the regression model obtained by nuclear correlation filter training is used to estimate the state of the next frame of the image,realize the automatic tracking of the status of Shenhua key accessories.The experimental results show that this method can obtain a relatively complete and clear state image of Shenhua key parts.The MCC values of different Shenhua key accessories are above 0.8,and the state detection results can be obtained before the abnormal state occurs.The status tracking results of Shenhua key accessories at each time are exactly the same as the actual results.