Research on metro traffic signal acquisition and analysis method based on image semantic recognition

At a time of rapid development in transport technology, metro has been widely used in urban public transport for short and medium distances, becoming an important component of the public transport network for the convenience of people's travel. Certain safety hazards may exist during the operation of the metro, however: intrusions from people and external construction can disrupt the operating environment and seriously affect the safety of trains during operation; deformation and settlement of the tunnel walls along the metro rails due to the passage of time and soil erosion beneath the tunnels will also have an impact on the continuous operation of the trains. At present, traditional safety inspection techniques remain largely manual with some semi-automatic means. Such methods are time-consuming and labour-intensive, and fail to provide timely and accurate warnings of possible safety hazards along the rail line. Therefore, it is of great significance to carry out real-time automated monitoring for the metro tunnel structure and along the metro rails. In this paper, a method of extracting vibration signals from the operation of metro rail trains using weak grating array fibers is discussed, and a method of analyzing demodulated signals based on image semantic recognition technology is pro- posed, which enables the monitoring of track operation status, track surroundings, personnel intrusion, etc.