Research On Low Frequency Ultrasonic Aliasing Signal Separation Method Based On Sparse Representation

Multilayer composite structural components are widely used in various engineering fields,and the detection of their thickness specifications and quality conditions has always been a hot research topic.Ultrasonic pulse reflection method is usually used to detect the thickness of multilayer composite structure,but there are multiple reflection echo overlapping and noise interference problems,which seriously affect the extraction of useful information of target echo.Therefore,the effective removal of noise interference and the separation of ultrasonic aliasing signals are very important to improve the accuracy of low-frequency ultrasonic testing of multilayer composite structures.In response to the above problems,this dissertation carries out the low-frequency ultrasonic detection experiment of large metal-nonmetal composite structures,and studies the noise reduction and separation methods of the obtained mixed signals.Firstly,based on the research of ultrasonic propagation theory and ultrasonic testing methods of multilayer composite structures,a low-frequency ultrasonic testing system for composite structures is built,and the ultrasonic testing tests are carried out on the tubular steel plate and rubber material bonded components,which verified the feasibility of using lowfrequency ultrasonic pulse reflection method to detect such components.Secondly,combined with the spectral characteristics of ultrasonic signals,a signal noise reduction algorithm based on variational modal decomposition is proposed in this dissertation to eliminate the interference of background noise on subsequent aliasing signal separation processing.The feasibility of the algorithm is verified by simulation signals,meanwhile,the algorithm is applied to the experimental signal processing of multilayer composite structure ultrasonic detection to analyze its noise reduction effect.Simulation and practical test results show that the algorithm can improve the signal-to-noise ratio of aliased signals and effectively preserve the signal details.Finally,based on Nakagami model,an over complete atomic dictionary that conforms to the characteristics of ultrasonic echo signal is designed,and the traditional orthogonal matching tracking algorithm is improved to improve the optimization accuracy of the algorithm.Through the simulation and actual testing of low-frequency ultrasonic aliasing signals,the algorithm improves the accuracy of echo approximation and the separation effect of aliasing signals.