Research on Intelligent Alarm System for Falling Rocks on Railway

Along with the rapid development of railway industry in China, the railway network becomes more complex. In order to ensure the train run with higher speed, more safety and stability, the FlowNet2.0 is used in our alarm system to avoid or reduce the damage of natural disasters and dangerous objects on train safety. FlowNet2.0 is based on traditional optical flow method, but it is optimized optical flow algorithm based on deep learning. We use it to develop a new intelligent alarm system for falling rocks along the railway. This system uses advanced video processing technology and the object’s contour, moving speed and angle in optical flow field to process the digital video captured in monitoring place in real time. When accidents, like rocks falling from revetment, landslide displacement, debris flow, foreign object intrusion, etc. happens on the rail track, this system can detect those disasters in real-time accurately, and alarm automatically. This system can replace the traditional watchman with low-cost equipment, detect disasters and alarm in real time. The deployment of this system can not only improve the safety of train significantly, but also save the labor cost from Railway Bureau.