| New fiber rigid ceramic matrix composites have been widely used in thermal protection fields such as aerospace,industrial equipment,and automobile industry due to their excellent properties such as light weight,high temperature resistance,low thermal conductivity,oxidation resistance,and corrosion resistance.The material is usually bonded to the substrate surface to provide thermal protection.However,during use,due to the bonding process and fatigue damage in the service process,bonding defects of the bonding layer will occur,which seriously threatens the operational safety of equipment and systems.Therefore,it is of great significance to carry out bonding defect detection and reliability evaluation of the ceramic matrix composite bonding members.In this paper,the method of detecting and characterizing the bonding defects of ceramic matrix composites based on ultrasonic C-scan was studied.The main contents include three aspects:numerical simulation of acoustic propagation characteristics of materials,time-frequency analysis of ultrasonic detection signals and quantitative characterization of defects.(1)Considering the complexity of acoustic properties of ceramic matrix composites,the propagation characteristics of ultrasonic in the composites have not been studied.In order to explore the influence of material properties on ultrasonic propagation characteristics,a two-dimensional transient propagation model of sound field was established.The propagation characteristics of ultrasonic in ceramic matrix composites were numerically simulated.The reliability of the simulation is verified by comparing the simulated sound velocity with the theoretical sound velocity.In this paper,the transient propagation of ultrasonic in the material at different times is studied,and then the acoustic field characteristics and acoustic attenuation in the material are simulated.The distribution law of ultrasonic sound field in the material and the law of energy attenuation in the process of ultrasonic propagation are revealed,which provides a reference method and theoretical basis for the study of the acoustic propagation characteristics of this kind of material.(2)Aiming at the problem that the porousness and anisotropy of the measured material cause the ultrasonic echo signal to be easily submerged by noise,and the signal characteristics are not easy to be extracted.This paper proposes to use wavelet packet time-frequency domain analysis and energy spectrum analysis to extract the spectrum and energy spectrum characteristics of the ultrasonic echo signal.Based on the analysis of the signal time-frequency characteristics,an adaptive wavelet packet threshold algorithm is used to enhance the echo signal.This feature effectively reduces the interference of noise signals to the original ultrasound signal.In order to realize the two-dimensional characterization of the ultrasonic C-scan image of the defect of the adhesive layer of the sample.In this paper,an ultrasonic C-scan detection system based on pulse reflection method is built independently.The detection of defects such as debonding,voids and inclusions of CMC material samples is carried out,and through comparative experiments,the adaptability of the detection method to different types of defects is analyzed.(3)In order to improve the accuracy of quantitative analysis of defects,a method combining convolutional neural network algorithm and Grab Cut image segmentation algorithm is proposed to achieve interaction-free segmentation and quantitative analysis of defects in ultrasonic C-scan images.The convolutional neural network is used to identify the defect in the ultrasonic C-scan image to determine the defect location.On this basis,the Grab Cut image segmentation algorithm is combined to realize the interaction-free segmentation and quantitative analysis of the defect image.The results show that this method can effectively identify and quantitatively analyze the defects,and achieve better quantitative accuracy of defects. |
PDF Full Text Request