Research And System Development On Safety Protection And Early Warning Technology Of Railway Unattended Crossing In Mining Area

Coal is the main energy source of our country and it takes up a dominant position in the structure of non-renewable energy consumption.With scientific and technological advance and social development,there is rapid needs in energy sources for our country.The railway in the mining is regarded the middle link of coal production,transportation and sales,it is of vital significance to ensure the safety of railway transportation.Compared with the nation’s railway,railway crossings in the mining area are numerous and few people supervise most of them.Pedestrians tend to ignore their vigilance when passing across these unsupervised railway crossings,leading to severe traffic accidents from time to time.Relative authorities have attached importance to the security issues.In order to ensure the transportation safety of vehicles at unsupervised railway crossings in the mining area,this research taken it as the application scenario,and adopts the technology of the wind-solar hybrid power supply at the crossing to solve the problem of no electric power supply at the unsupervised crossings.The technology realizes the detection and early warning of road conditions at the unsupervised crossings.The contents of the specific research include the following aspects.Firstly,in view of the lack of reliable power supply at the unattended crossing of the railway in the mining area,the sustainable power supply mode of the intelligent security robot at the crossing was carried out.Design the wind-solar complementary power supply system to power the intelligent security robot at the crossing,increase the battery in the power supply system for energy storage,and improve the reliability of the intelligent security robot power supply at the crossing.Secondly,the research aims at targets,such as pedestrians,vehicles and other targets at the crossing.In order to improve the detection accuracy,the author designed a DIY’s data set on the basis of the BDD100 K data set.In order to increase the number of the sample,the author applies supervised data augmentation and data augmentation based on improved DCGAN,and establishes a data set that meets the traffic characteristics of the site through screening.Thirdly,according to the complex characteristics of railway crossing vehicles in mining area,the author makes a analysis and comparison for several target detection algorithms at present and is determined that target detection is based on the YOLOv5 algorithm.The author applies the technology of the label smoothing regularization as a solution to the over-fit phenomena in the experiment process.Based on the uneven illumination in the actual detection,the fusion convolution attention model is used for improvement.And the Bi FPN structure is adopted in the Neck layer to improve the fusion of target multi-size features.The improved algorithm is experimentally analyzed with the DIY’s data set,and the effectiveness of the algorithm improvement is verified.The last step is that the application system design is carried out for the security warning system of the unsupervised crossing in the mining area,and the installation and debugging are carried out.The system includes a crossing intelligent security robot,a vehicle-mounted system and a host computer monitoring system.The application test of the built intelligent security early warning system is carried out,and the test data is analyzed to further verify the effectiveness of the wind-solar complementary power supply system and the optimized YOLOv5 target detection algorithm.