Research On Recurrence Analysis Of Ultrasonic Backscattered Signal And Its Application In Defect Detection Of Carbon Fiber Reinforced Polymer

With advances in manufacturing technology and requirements of structural performance,Carbon Fiber Reinforced Polymer(CFRP)has shown a trend of complex stacking structure,multiple layers and large thickness.The manufacturing process of thick section CFRP is complicated,and the problems of temperature gradient and residual stress often lead to manufacturing defects such as uneven fiber distribution,resin-rich,and fiber waviness.Defects in thick-section CFRP components occur randomly in different locations and sizes,while early micro-defects may develop into serious defects in the service process of components,and distal defects of large components can not be detected because they are far away from the detection surface.New requirements have been put forward for the detection and characterization of defects in CFRP.Ultrasonic testing is one of the most widely used non-destructive testing methods,which can be realized by different devices.The pulse reflection method that uses single side probe for both transmission and reception is the most direct and practical ultrasonic testing method.The signal obtained by the pulse reflection method contains rich information on the propagation behavior of the ultrasonic pulse in the material,and each part of the signal has been effectively applied to the defect detection of CFRP:defect detection based on energy change of back-wall echo,pulse resonance behavior in near-surface backscattered signal,abnormal energy component in backscattered signal,etc.However,the application of pulse reflection methods is limited for small defects in thick-section CFRP that are located far from the detection surface.For example,the resonance frequency component of the far-side backscattered signal almost disappears,and the attenuation of the ultrasonic pulse increases with the increase of the propagation distance,while the back-wall echo may even disappears completely.Besides,the inherent non-uniformity of CFRP will also form abnormal reflection echoes in the backscattered signal.All these facts show that defect detection method based on signal energy is no longer efficient in detection of far-side micro defects in thick section CFRP,and it is necessary to consider the defect characterization method from a new perspective.Time series analysis studies the evolution law of the system based on the relationship between the data,and has low requirements on the absolute amplitude of the signal.It can be used for the analysis of far-side backscattered signals.The linear time series analysis model is simple and easy to implement,and has a good predictive effect on stationary signals.However,due to the complicated lay-up of CFRP and the complex propagation behavior of ultrasonic pulses,the statistical characteristics of the ultrasonic signal change with time,so that the ultasonic signal acquired is nonstationary.Moreover,the length of time series acquired is relatively short.The stationarity assumption and data requirement of linear time series analysis can not be satisfied,so it is not suitable for ultrasonic signal analysis of thick section CFRP.Nonlinear time series analysis methods such as recurrence analysis obtain system evolution rules by studying the recurrence behavior of trajectories in phase space.It is free from the limitation of the stationary assumption,and can analyze the short-time non-stationary signal,and thus feasible for detection of far-side micro defect in thick section CFRP.Therefore,in this paper,the pulse reflection method was used to detect the far-side micro defects,and the ultrasonic backscattered signals are analyzed by the nonlinear time series methods to improve the detection ability of the traditional ultrasonic testing.The contents of this article mainly include the following aspects.:(1)Introduce the application scenarios and research status of CFRP,analyze the advantages and disadvantages of traditional ultrasonic testing methods and advanced ultrasonic testing techniques,and discuss the feasibility of nonlinear time series analysis of backscattered signals.(2)Perform a linear model analysis on the pulse reflection process,analyze the formation process,composition and linear characteristics of the backscatter signal,and study the influence of CFRP internal structure and defects on the backscatter signal.(3)Explain the theory of nonlinear time series analysis,introduce the theory of phase space reconstruction,and analyze the phase space of CFRP backscattered signals with different thicknesses to obtain the trajectory characteristics of backscattered signals.The relationship between the optimal reconstruction parameters and the material structure was also analyzed.(4)Introduce the analysis tool based on phase space reconstruction-recurrence analysis method,including recurrence plot(RP)and recurrence quantification analysis(RQA).By studying the RPs of different thickness materials,the selection of the corresponding threshold parameters was determined and the structural characteristics of the backscattered signals were qualitatively recognized.The RQA variables of the backscattered signals of each CFRP sample were obtained to quantify the nonlinearity of the backscattered signals.The results showed that the far-side backscattered signal of thick-section CFRP has great instability.(5)For the instability of far-side backscattered signal,the long-term behavior of the ultrasonic signal was statistically analyzed and described using statistical RQA values.In the comparison experiment,for the same thick section composite material,repeated measurement in the defect-free area and the defective area were conducted respectively and several sets of signals were obtained.The RQA analysis was performed in each units of groups.Statistics of RQA variables of each group can be used to obtain the characterization of signal instability:RQA median and interquartile range.With the comparative analysis of the statistical RQA value of the defect-free and defective areas,the influence of the defect was known and the existence of the artificial artifact can be effectively determined.(6)Experimental research on the relationship between ultrasonic testing equipment parameters and far-side backscattered signals have been performed.Local trajectory prediction denoising method was proposed for the denoise of the signal.Thick CFRP was detected by using acquisition devices of different sampling frequency and ultrasonic probes of different center frequencies,to study the difference of phase space reconstruction parameters of backscattered signals obtained.The backscattered signal was processed by the time domain averaging method,the wavelet threshold denoising method and the local trajectory prediction based denoising method.The local trajectory prediction denoising method and the wavelet threshold denoising method can effectively remove the noise and restore the system structure.However,the two methods have different emphasis on the removal of backscattered signal components.(7)At the end of the text,the paper summarizes the full text,analyzes the shortcomings of the proposed method,and forecasts the possible future research directions.