Nonlinear Analysis And Quality Evaluation Of C_f/PA6 Ultrasonic Welding Based On Vibration Signal

Lightweight automobile structure can effectively reduce energy consumption and carbon dioxide emissions.It is one of the important measures to realize the national strategy of "Carbon Peak" and "Carbon Neutrality" as well as the sustainable development of human society.The lightweight of automobile structure is mainly based on new reinforced composite materials or alloy materials instead of traditional metal materials used in automobiles.Large or complex reinforced composite components are mainly connected by using ultrasonic welding technology.The quality of welding directly determines the service quality of the final composite material.Therefore,welding quality monitoring has always been the focus of research.At present,the ultrasonic nondestructive testing of welding quality mainly is mainly based on postwelding quality testing.This method cannot obtain timely welding feedback signals,and it is not convenient to repair the welding quality.Therefore,the online real-time detection of ultrasonic welding is highly required.Additionally,it can also be employed to combine with artificial intelligence to achieve the ultimate goal "Intelligent Testing".In this thesis,the carbon fiber reinforced nylon 6(Cf/PA6)composite materials were chosen as samples and investigated by using ultrasonic welding technique.The ultrasonic vibration signals from the upper and lower plates in the welding process were collected in real time and the nonlinear characteristics of vibration signals were analyzed.The spectrum energy difference between the upper and lower plates of the weldment was extracted as the characteristic information.This information was employed to evaluate the welding strength based on Support Vector Regression(SVR)algorithm,thus providing the possibility for online real-time testing by using ultrasonic vibration signals.The main contents of this thesis are summarized as follows:1.Firstly,the relationship between acoustic nonlinearity and material nonlinearity in solid medium is introduced.Then,the cause of high-order harmonic and the principle of ultrasonic nonlinearity detection are explained.Furthermore,the method of solving the second and third nonlinearity parameters is also explained.2.In view of the experimental environment of ultrasonic welding and the characteristics of ultrasonic vibration signals,an ultrasonic vibration signal acquisition system is designed.Additionally,in order to facilitate the processing and analysis of vibration signals,a graphical software for vibration signal processing has also been designed and developed based on MATLAB.3.By using different welding pressure to obtain the welded joints with the different strength,the correlation between nonlinear deformation of joint and joint strength was analyzed.Based on the ultrasonic transmission model in the welding process,the harmonic components of the vibration signals from the upper and lower plates were analyzed.Based on the variation signals harmonic amplitude,the second and third nonlinearity parameters of the upper and lower plates were calculated,and the nonlinearity parameter under different welding pressures were also explored.The results show that both the vibration signals of the upper and lower plates are affected by the deformation of joint.Both the signals of upper and lower plates deliver abundant acoustic nonlinear information,which provides an effective reference for real-time online monitoring of welding quality.4.In order to analyze the nonlinearity of vibration signals in time-frequency,the results based on the three different time-frequency spectra(i.e.,Short-Time Fourier Transform,Wavelet Transform and Wavelet Packet Transform)were compared.The result based on the wavelet packet transform can more accurately reflect the characteristics of nonlinearity in the welding process.Then,by using the Wavelet Packet Transform,the vibration signals were decomposed into three layers.The obtained wavelet packet energy spectra effectively extract the nonlinear characteristics of the vibration signals of the upper and lower plates.5.The wavelet packet energy spectrum difference between the upper and lower plate signals was selected as the characteristic information,and the SVR and BP neural network were used to perform regression prediction on the joint strength.The results show that SVR has better prediction ability with a high determination coefficient(0.9205),a mean square error(0.0172)and mean absolute error(0.1713),indicating that the quality of ultrasonic welded joint can be well evaluated by using the nonlinear characteristics of ultrasonic signal.In this thesis,the ultrasonic welding process and welding quality of Cf/PA6 composites was studied in terms of vibration signals,material nonlinearity and ultrasonic nonlinearity,and the welding quality were evaluated combined with"artificial intelligence".It is expected that the findings in this thesis can provide useful guidance on the future development of practical online real-time monitoring of ultrasonic welding quality,and accelerate the progress of automobile lightweight design,energy saving and emission reduction.