The Research On Defect Detection Of CFRP Reinforced Concrete Based On Infrared Thermal Imaging

Concrete is often used as the main load-bearing structure in construction,but in the process of construction,due to improper operation of workers or the influence of external environment and other factors,it may lead to internal defects of the structure.At present,the detection technology for structural internal defects is not mature enough.How to realize non-contact,high-precision defect detection and depth inversion is a key problem.In this paper,active infrared thermal imaging technology is applied to defect detection and quantitative inversion of CFRP reinforced concrete.The main contents of this paper are as follows:(1)Existing infrared defect detection algorithms can only achieve defect location on the surface or shallow surface of the structure,but fail to achieve quantitative inversion of defect depth.In this paper,combined with the basic theory of heat transfer,the defects in reinforced concrete structures are studied,and a mathematical model for calculating the buried depth of heat insulation defects and the thickness of heat conduction defects is established.(2)By using the established mathematical model,the defect depth and thickness of simulation samples are preliminarily inverted.The preliminary inversion results show that the inversion error of the center burial depth of the heat insulation defects is within 6%,and the inversion error of the thickness of the heat conduction defects is within 7%.The reliability of the mathematical model is proved.(3)Aiming at the influence of radial thermal diffusion on inversion results,an inversion error compensation model based on XGBoost is proposed in this paper.At the same time,in order to further improve the inversion accuracy,centroid prior knowledge and distance prior knowledge are applied,and simulation samples are used to train the model,so as to realize the correction of burial depth inversion error of simulation samples.(4)An experimental platform of active infrared thermal imaging defect detection was established to conduct qualitative and quantitative experimental research on the defects in reinforced concrete structures.Principal component analysis algorithm was used to remove the influence of uneven heating on the thermal image,and based on the low contrast between infrared image target and background,the threshold segmentation algorithm combined with genetic algorithm and maximum entropy method was used to complete the segmentation and defect extraction of infrared image.On this basis,the inversion of the defect depth and thickness of the actual sample is completed by using the established mathematical model.In view of the difficulty in acquiring measured samples and the small number of measured samples is far from enough for the training error compensation model,this paper uses a small number of measured samples to fine-tune the error compensation model based on simulation sample training,and realizes the correction of the actual sample defect burial depth inversion error,thus improving the inversion accuracy of the overall defect region.