Infrared Nondestructive Testing Of Composite Materials Based On Segmentation Network

Most common types of defects for composite are debond and delamination.The Optical Pulsed Thermography(OPT)technology has the advantages of fast detection speed,large detection area and easy operation.It is a non-destructive testing technology suitable for composite materials,and plays an important role in aerospace and military fields.However,the original detection results will be interfered by factors such as edge information,background,noise,etc.It is necessary to use different algorithms to improve the signal-to-noise(SNR)ratio.At present,domestic and foreign scholars have proposed many different detection algorithms to extract the features of defects and increase the SNR of infrared thermography images,but there are still some limitations,such as,the detection accuracy and the detection rate of internal defects on complex shape specimens needs to be improved.In view of the above problems,and considering the outstanding effects of segmentation networks in the field of natural image processing,this paper combines the infrared nondestructive testing and deep learning.After analyzing and comparing the temporal-based and spatial-based segmentation network models,a hybrid of spatial and temporal deep learning architecture for automatic thermography defects detection is proposed.The integration of cross network learning strategy has the capability to significantly minimize the uneven illumination and enhance the detection rate.The main research work of this paper is as follows:(1)An OPT nondestructive testing system was built,and four types of defective test pieces with different diameters and depths were detected using the system.The original data obtained is analyzed and found to be decomposed into spatial mode and temporal mode,thus leading to the exploration of combining the basic principles of infrared thermal imaging with deep learning.In order to enhance the contrast between the defect area and the sound area and reduce the influence of noise,the commonly used defect extraction algorithm(Principal Component Analysis and Thermographic Signal Reconstruction)are used to process the obtained data,and the advantages and disadvantages of different methods are analyzed and compared.(2)By analyzing the thermal spatial characteristic and temporal characteristic of thermal video data in deep learning structure,the corresponding data sets are established according to the data characteristics and a temporal and spatial deep learning network for infrared thermal defect detection is proposed.The method can be applied to the automatic segmentation of composite defect detection by OPT system.In order to verify the stability and validity of the proposed method,the internal debonding defects of conventional and complex shape composite specimens were segmented by the method and the probability of detection(POD)has been derived to measure the detection results.The results show that VGG-Unet cross learning structure can significantly improve the contrast between the defective and non-defective regions.In addition,in this paper,for specimens with low SNR and deeper defect depth,the PCA is integrated into VGG-Unet to optimize the learning structure.