Design And Implementation Of Trackside Equipment Detection System For High-Speed Railway

With the rapid development of economy and science and technology,the high-speed railway has developed rapidly nowadays.China’s highspeed railway system has reached 35,000 kilometers(2019)in total mileage,ranking first in the world.In addition to the most basic track devices,the safe operation of trains on high-speed railways is also crucial for the equipment next to the tracks.Various signals on high-speed railways are transmitted by signal boxes near the rails.Regular inspection of trackside equipment is also particularly important.Based on the actual conditions and detection methods of high-speed railway trackside systems,this paper designs and develops a system that can automatically identify and detect trackside equipment on the trackside.The image information of relevant trackside equipment can be directly obtained from the images collected in the high-speed railway track inspection vehicle.The convolutional neural network model is used as the main training part,and basic image processing techniques such as vertical projection are used for preliminary image screening.Image data annotation and transfer learning,and finally analyze the experimental results and select the final model through different network models and different parameters.It is mainly based on the existing collected railroad trackside image data to initially screen out possible images as a data set,and use convolutional neural network and open source software framework Tensor Flow and Object Detection API for deep learning,and based on this Generating a target object detection model is of great significance for improving the detection and routine maintenance of high-speed railway trackside equipment and reducing labor costs.Through experiments,it is confirmed that the accuracy and rate of recognition of target equipment next to the track by the convolutional neural network in this paper are relatively good,which achieves the main purpose of this study and has certain application value in the field of high-speed railway trackside equipment detection.