Research On Feature Extraction And Imaging Method Using Laser Ultrasonic Of Laminated Composites

Composite materials are widely used in aerospace,medical equipment,automotive industry and other fields.Due to the limitations of the material preparation process,defects such as delamination,debonding,and inclusions may occur during the preparation of composite materials,affecting the performance and production safety of the material.As an emerging detection technology,laser ultrasound has the characteristics of non-contact,high precision and complex surface detection,and can complete the performance analysis and defect detection of composite materials.In this paper,the use of laser ultrasonic detection of laminated composite materials,the laser ultrasonic signal is noisy,difficult to extract features and imaging results are not easy to segment the problem of feature extraction and imaging methods.In this paper,the laser ultrasonic detection method of laminated composite materials is analyzed.The ultrasonic C-Scan signal of the composite material bonding plate is collected by transmission method,and the time-frequency analysis and imaging analysis of the ultrasonic signals are performed.The dbN,symN and The feature extraction effect of coifN wavelet family under different decomposition layers and different threshold processing functions,using db2 wavelet to decompose the echo signal,soft threshold processing and reconstruction,improves the signal-to-noise ratio of the laser ultrasonic signal,and effectively extracts Ultrasonic feature signal reflecting the information of layered defects;for the problem that the defects caused by the peaks and valleys of image pixels are difficult to segment,the image pre-processing is performed using the 4-neighbor mean processing method;based on the histogram method and Otsu method Image segmentation technique of maximum between-class variance method weighted by the sum of neighborhood gray-levelprobability.Through the quantitative analysis of the defect size,the maximum inter-class variance method weighted by the sum of 28 neighborhood gray scale probabilities was used to segment the defect image,and the accuracy rate was as high as 96.1%.