Research On Some Key Technologies For Ultrasonic Testing Of Carbon Fiber Reinforced Plastics

The carbon fiber reinforced plastic(CFRP)has been widely used in aircraft industry due to its high strength and low density.Unfortunately,flaws in CFRP will degrade the performance.Consequently,strict quality control in CFRP fabrication is essential.The nondestructive testing(NDT)technologies have played an important role in quality control and the ultrasonic inspection technology has been the one mostly extensively used.In this dissertation,inspection of flaws in CFRP based on ultrasonic technology is investigated and three aspects,i.e.,the modelling of propagation of ultrasonic wave in CFRP,the porosity evaluation,the localised porosity detection and the large-size flaws detection,are mainly involved.The details are given as follows.(1)The propagation of ultrasonic wave in CFRP is modelled.Certain models about the propagation of ultrasonic wave in multilayered structures are introduced,and the resonance in CFRP is investigated.Meanwhile,the influences of fiber volume fraction,nominal ply thickness,the voids,and the rich-resin on reflection coefficient are discussed.Furthermore,the waveform of received signal is simulated,and the time-frequency(TF)distributions in the cases of unflawed CFRP,CFRP containing void and CFRP containing rich-resin are studied.An approximate model is subsequently established to give a sparse representation of the received signal and reveal the multicomponent characteristic of the received signal.(2)The application of nonlinear dynamic methods in porosity evaluation is investigated.The recurrence plot(RP),the recurrence qualification analysis(RQA),and the multiple-scale sample entropy(multi-scale SE,MSE)versus porosity are investigated,providing effective tools for porosity evaluation of CFRP workpieces from the same batch.The effects of parameters selection,distance criteria,and scales are also investigated.The results indicate that the nonlinear dynamic methods are promising techniques in porosity assessment in the abscence of back-wall echo,which often occurs when the tested workpieces are thick or of complex geometries.(3)The detection of localised porosity in low-porosity CFRP is studied.For the low-porosity CFRP,the possible localised porosity will degrade the performance despite of qualified porosity.Consequently,detection of localised porosity in low-porosity CFRP is essential.The variational mode decomposition(VMD)is introduced at the first stage.The methods for parameters estimation of ultrasonic echoes formulated by the Gaussian chirplet model,including the Gaussian-Newton(GN)algorithm,the expectation maximization(EM)algorithm are studied.The matching pursuit(MP)is introduced subsequently.A preliminary method for localised porosity detection based on VMD and ultrasonic echoes parameters estimation methods is proposed.To begin with,the VMD is used to separated the mixed noise,the 1st resonance signal,and the low-frequency component containing the flaw echoes from the received signal.Subsequently,the echoes in the extracted low-frequency component are separated based on the parameter estimation.Finally,the concepts of energy function and instant gains are put forward,and the reflectivity of plies where those flaw echoes locate are evaluated,yielding an approximately quantitive assessment of localised porosity.(4)The detection methods of large-size flaws in CFRP with complex geometry are put forward.Large-size flaws,such as void close to the millimeter level,delimination,and slender crack are possible in those CFRP with complex geometries in the process of manufacturing,interfering the work of porosity evaluation.More seriously,the performance of those CFRP containing large-size flaws will be degraded significantly.As a result,detection of large-size flaws in CFRP with complex geometry is necessary.Firstly,the TF distributions of ultrasonic echoes produced by the generalized S-transform are analyzed.Method for echoes recovery based on TF image processing is proposed.Three concepts are associated with the proposed method,which are the adaptive threshold,the morphylogy,and the consecutive-difference based measurements.Secondly,the empirical mode decomposition(EMD)-based denoising strategies are introduced,and the SE of all intrinsic mode functions(IMFs)are analyzed.An EMD-based denoising method on the basis of IMF-SE based partial reconstruction and the window processing based coherent noise suppression is proposed.Finally,the basic theories of support vector machine(SVM)and some classification performance measurements are introduced,and a SVM-based denoising method is proposed.(5)The visualization techniques of ultrasonic testing results are proposed.The selection of visualization object,the alleviation of redundancy and "shadowing",and the generation of multiple views are investigated for quick identification of defects.A preliminary ultrasonic testing system has been developed to meet the current demand of ultrasonic signal processing and visualization of testing results.