Fault Detection of Train Height Valve Based on Nanodet-Resnet101

The train height valve can adjust the charging and deflating of the air spring according to the trainload. The normal operation of the height valve ensures the smooth operation of the train and prevents the track from being damaged due to wheel impact. Therefore, the maintenance of the height valve is very important. Traditional manual maintenance consumes a lot of manpower and material resources, at the same time, efficiency and accuracy are difficult to be guaranteed. In this paper, a fault detection algorithm based on Nanodet-Resnet101 is proposed. Through experiment, the network has a 97 percent accuracy rate in simultaneously detecting an image in 0.1 seconds. The experimental results show that the height valve fault detection network combined with NanoDet and Resnet101 has engineering significance