Research on Intelligent Inspection Vehicle for Suspended Air-Rail Track Beam Based on Multi-Sensor Fusion and State Recognition

Due to the complex structure of suspended air rail track beams, they are easy to wear and corrode under the action of load and natural environment, and the traditional manual flaw detection exists problems such as many blind zones, low precision, high altitude risk and so on, which is difficult to meet the demand of efficient and accurate detection. This project develops an intelligent inspection vehicle based on multi-sensor information fusion and state recognition, adopting IMU inertial sensor fusion technology, SCNN network, YOLOv5s6 algorithm, and Xception network framework, which respectively improve the processing capability of inspection elements, the track fitting accuracy, the precision of detecting small objects, and the meticulousness of track state assessment. Through multi-source data fusion and algorithms, the complex working condition position calibration is realised, the detection accuracy and recall rate are improved, and the damage knowledge base is established to realise the state assessment and damage dynamic prediction. After on-rail experiments, the detection accuracy is high, the coverage is comprehensive, and it adapts to complex conditions, making the detection scientific, standardised, comprehensive and efficient.