Intelligent Identification Method for Subgrade Layer Deformation of Ballasted Railway
In response to the challenge of acquiring subgrade layer changes in single-phase ground-penetrating radar(GPR)data,an intelligent identification method for subgrade layer deformation based on periodic detection has been proposed.Firstly,the YOLO v5 model was employed to identify bridge equipment in radar images.By matching with the equipment table,the mileage registration of multi-temporal data was achieved.Subsequently,the U-Net model was utilized to accurately identify the subgrade location lines in multi-phase data.Finally,based on the annual deformation,the mileage range of significant subgrade deformation was extracted,providing data support for maintenance and repair decisions.Subsequently,experimental tests were conducted using actual measured data.The results indicate that the registration accuracy of multi-phase data meets the application requirements.The automatically identified layer boundaries are close to the results of manual tracing,effectively enhancing the processing efficiency and precision of periodic ground penetrating radar(GPR)detection data.This could provide a novel approach for detecting deformation in railway subgrade layers.
万方数据
维普
