Research On Cable Force Identification And Sheath Defect Detection Technology Of Bridge Cable

The cable is an important component of modern long-span structural bridge,so it is very important to master its operating state to ensure the safety of the bridge.Bridge cable is mainly composed of cable body and outer PE protective sheath.This paper focuses on two aspects of bridge cable force identification and sheath surface defect detection,and studies the cable force identification method based on spectrum analysis and the cable sheath surface defect detection method based on deep learning,so as to provide certain technical support for realizing the informatization and intellectualization of bridge cable health status assessment.The main research contents are as follows:(1)The difficulty of identifying multi-component time-varying modal parameters in the vibration displacement signals of bridge cable is studied,and a new structural instantaneous frequency identification method based on variational mode decomposition(VMD),wavelet analysis and isoline method is proposed.Firstly,the number of modal components is determined by the continuous wavelet transform,and the multi-component signals are decomposed by VMD to obtain each modal component.Then,the instantaneous frequency of each component signal is identified by wavelet analysis and isoline method.Finally,the instantaneous cable force is calculated according to the "Cable force-Frequency" formula.Wavelet exponential function threshold denoising method is introduced to reduce noise interference.By means of signal splitting and time period analysis,a more appropriate denoising threshold can be determined and the problem of limited support length of constant amplitude surface can be solved.Numerical examples and time-varying stiffness cable tests are designed to verify the results.The results show that the proposed time-frequency analysis method can identify the instantaneous frequency of time-varying structures accurately and effectively,and has strong anti-noise performance.Based on the visual interface development platform provided by Matlab,the GUI interactive interface of cable force identification and analysis system is designed.The program modularization,visualization,lower the user threshold.(2)The image quality and defect characteristics of the bridge cable sheath surface were analyzed,and the defects were divided into three types:mild damage type,moderate damage type and severe damage type according to the damage degree of the defects to the sheath.To solve the problem of uneven illumination,MSR algorithm is used to enhance image.In order to alleviate the problem of small samples with defective images,the data augmentation technique is used to expand the sample size of data set.The U-Net semantic segmentation network suitable for small sample data sets was selected to segment the surface defects of the sheathing,and the network was improved.VGG16 network skeleton was used to replace the encoder,CBAM module was added before the pooling layer to improve the segmentation accuracy of the small target model.The Focal loss+Dice loss function with weight was used to overcome the unbalance between samples.The model is trained and tested on the laboratory computer,and the experiment shows that the segmentation performance of the improved U-Net model is greatly improved,reaching 85.36%mIoU,93.35%mPA and 99.0%Acc.In the video detection,the FPS is about 6.3.Through the above research,the task of dynamic displacement radar sensor dynamic measurement of time-varying cable forces is basically completed.Through the above research,the task of dynamic displacement radar sensor dynamic measurement of time-varying cable forces is basically completed.By improving and training U-Net network,an effective video detection model of cable sheath surface defects is obtained,which is expected to realize the automatic detection function of cable climbing robot.This paper is of great practical value for building bridge intelligent monitoring BIM system.By improving and training U-Net network,an effective video detection model of cable sheath surface defects is obtained,which is expected to realize the automatic detection function of cable climbing robot.This paper is of great practical value for building bridge intelligent monitoring BIM system.