Dynamic detection method for weak bonding defects of CFRP interface based on peak entropy

Abstract

The application discloses a CFRP interface weak bonding defect dynamic detection method based on peak entropy, and specifically comprises the following steps: step 1, collecting particle velocity signals on the back of a nondestructive test piece under laser impact through a photonic Doppler velocimetry system; step 2, pre-processing the particle velocity signals; step 3, performing S transform time-frequency analysis and extreme value analysis on the pre-processed particle velocity signals, and calculating amplitude differences between extreme value points; step 4, calculating Shannon entropy of the amplitude difference sequence between the extreme value points, and generating a peak entropy feature; and step 5, extracting the peak entropy feature, and judging a damage state of the nondestructive test piece. The CFRP interface weak bonding defect dynamic detection method based on the peak entropy collects dynamic response signals through single laser impact, adopts S transform time-frequency analysis and Shannon entropy feature extraction, and realizes single impact detection.

Description

Technical Field

[0001] This invention belongs to the technical field of nondestructive testing methods for composite materials, specifically relating to a dynamic detection method for weak bonding defects at CFRP interfaces based on "peak entropy". Background Technology

[0002] Carbon fiber reinforced polymer (CFRP) composites are widely used in the aerospace field due to their lightweight and high specific strength properties. However, their adhesive interfaces are prone to "weak bonding defects" caused by manufacturing or service issues. Traditional non-destructive testing (NDT) techniques, such as ultrasonic and infrared thermography, rely on physical or chemical methods to detect internal or surface defects in materials, primarily depending on changes in the material's geometric morphology or differences in acoustic impedance. However, for weak bonding defects at CFRP interfaces, such as bonding strength below 20% of normal values, traditional testing techniques typically do not produce significant changes in geometric morphology (e.g., cracks, pores) or significant differences in acoustic impedance. Therefore, traditional techniques like ultrasonic and infrared thermography struggle to detect these defects. Existing entropy methods, including permutation entropy and fuzzy entropy, while capable of characterizing signal disorder, suffer from the following limitations: they cannot distinguish between material non-uniform noise and entropy differences caused by localized damage; they rely on signal analysis under quasi-static loading, making it difficult to capture dynamic impact transient responses; and they require extensive labeled data, resulting in insufficient diagnostic accuracy under small sample conditions. Laser shock wave interface bonding strength testing (LBI) technology activates defects through the stress wave reflection tensile effect, but existing methods still suffer from problems such as low efficiency and high cost.

[0003] LSPT, a US company, employs a three-stage impact testing method: a low-energy initial test to calibrate the reference signal, a high-energy test to activate defects, and a low-energy test to verify damage. This method relies on a fixed-pulse-width laser, requires multiple impacts to cover interfaces at different depths, and depends on complex signal comparisons, resulting in long testing cycles and significant equipment wear. The Air Force Engineering University has proposed a two-stage impact testing method: a low-energy initial test to calibrate the reference signal, a high-energy test to activate potential weak bonding defects, and comparison of the stress wave signals from the two impacts to determine if delamination exists at the bonding interface. This method requires high-precision extraction of dynamic response features and is sensitive to noise. However, there are still shortcomings in the analysis of signal coupling patterns in complex multi-interface structures. Summary of the Invention

[0004] The purpose of this invention is to provide a dynamic detection method for weak bonding defects at CFRP interfaces based on "peak entropy". The method acquires dynamic response signals through a single laser shock, and uses S-transform time-frequency analysis and Shannon entropy feature extraction to achieve single-shock detection.

[0005] The technical solution adopted in this invention is a dynamic detection method for weak bonding defects at CFRP interfaces based on "peak entropy", which specifically includes the following steps:

[0006] Step 1: Collect particle velocity signals on the back of the non-destructive specimen under laser shock using a photon Doppler velocimetry system;

[0007] Step 2: Preprocess the particle velocity signal;

[0008] Step 3: Perform S-transform time-frequency analysis and extreme value analysis on the preprocessed particle velocity signal to calculate the amplitude difference between extreme points;

[0009] Step 4: Calculate the Shannon entropy of the amplitude difference sequence between extreme points to generate peak entropy features;

[0010] Step 5: Extract peak entropy features to determine the damage state of the non-destructive specimen.

[0011] The invention is further characterized in that,

[0012] In step 1, carbon fiber reinforced composite material was selected as the non-destructive test specimen.

[0013] Step 2 specifically involves: first, downsampling the original signal sampling rate from 20 GS / s to 1 GS / s, then using baseline drift correction, extracting the low-frequency trend term using wavelet decomposition, and eliminating the signal baseline offset using algebraic subtraction; finally, using a low-pass digital filter with a cutoff frequency of 500 MHz to eliminate environmental noise and high-frequency interference from the instrument.

[0014] Step 3 specifically involves: performing S-transform time-domain analysis on the preprocessed signal to locate the transient energy peak region caused by stress wave reflection-unloading coupling; then identifying the peaks and troughs of the particle velocity curve using the second-order difference method, and eliminating spurious peaks with adjacent extreme value intervals less than 0.8 μs; taking the first main peak as the starting point, selecting the time interval from 20% to 30% of the velocity decay to the peak value as the feature analysis window, and statistically analyzing the probability distribution of the amplitude difference between consecutive extreme points within this time interval.

[0015] Step 3, which uses the second-order difference method to identify the peaks and troughs of the particle velocity curve, specifically involves:

[0016] First, the first-order difference is calculated on the denoised velocity curve to obtain the rate of change of the particle velocity signal; then, the first-order difference data is subjected to another difference operation to obtain the second-order difference, which reflects the curvature change of the particle velocity signal, as shown in formula (1):

[0017] (1)

[0018] In the formula, It is the particle velocity signal at point The value at; h It's the step length.

[0019] Step 3 involves calculating the amplitude difference between extreme points as follows:

[0020] (2)

[0021] In the formula, and These represent adjacent extreme points, Indicates the amplitude difference. .

[0022] The probability distribution of amplitude differences between consecutive extreme points within the time interval of step 3 is statistically analyzed using an adaptive binning method.

[0023] Step 4 is as follows:

[0024] The peak entropy value is calculated using the Shannon entropy formula, reflecting the degree of disorder in the extreme value changes of the particle velocity signal, as shown below:

[0025] (3)

[0026] In the formula, Representing the The probability of a difference in amplitude.

[0027] Step 5 specifically involves:

[0028] When peak entropy When a step occurs, if the peak entropy of the undamaged specimen is "0", the signal disorder is low and the specimen is in an undamaged state; if the peak entropy is greater than "0", the signal disorder increases and the specimen is in a damaged state.

[0029] Compared with the prior art, the beneficial effects of the present invention are:

[0030] (1) The dynamic detection method for weak bonding defects at CFRP interfaces based on "peak entropy" provided by this invention achieves a quantitative description of the degree of signal disorder by statistically analyzing the probability distribution of amplitude differences between consecutive extreme values ​​in the signal. Detection can be completed with a single impact, avoiding the cumbersome process of three impacts ("low-high-low") or two impacts ("low-high"); the peak entropy feature is sensitive to interface damage, resulting in high diagnostic accuracy; and it reduces laser consumption and experimental time, thus lowering costs. Attached Figure Description

[0031] Figure 1 This is a flowchart illustrating the dynamic detection method for weak bonding defects at CFRP interfaces based on "peak entropy" according to the present invention.

[0032] Figure 2This is a schematic diagram of the particle velocity curve on the back side of a specimen with a thickness of 1.5 mm in Embodiment 7 of the present invention;

[0033] Figure 3 This is a schematic diagram of the particle velocity curve on the back side of a specimen with a thickness of 3 mm in Embodiment 7 of the present invention;

[0034] Figure 4 This is a trend diagram showing the influence of peak entropy on the specimen under different impact parameters in the undamaged and damaged states in Embodiment 7 of the present invention. Detailed Implementation

[0035] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. The described embodiments are only some embodiments of the present invention, and not all embodiments.

[0036] Example 1

[0037] This invention provides a dynamic detection method for weak bonding defects at CFRP interfaces based on "peak entropy," such as... Figure 1 As shown, specifically:

[0038] Step 1: Collect particle velocity signals on the back side of a carbon fiber reinforced polymer (CFRP) specimen under laser shock using a photon Doppler velocimetry (PDV) system.

[0039] Step 2: Preprocess the particle velocity signal:

[0040] Step 3: Perform S-transform time-frequency analysis and extreme value analysis on the preprocessed particle velocity signal to calculate the amplitude difference between extreme points;

[0041] Step 4: Calculate the Shannon entropy of the amplitude difference sequence between extreme points to generate peak entropy features;

[0042] Step 5: Extract peak entropy features to determine the damage state of the non-destructive specimen.

[0043] Example 2

[0044] Based on Example 1, step 2 specifically involves:

[0045] The sampling rate can be reduced from 20 GS / s to 1 GS / s by downsampling, but the Nyquist sampling theorem must be satisfied. As long as the sampling rate is not less than twice the highest frequency in the signal, the true shape of the signal can still be restored after reducing the sampling rate without distortion.

[0046] This method reduces the amount of data to 1 / 20 of the original, resulting in faster subsequent processing while preserving the key characteristics of the signal.

[0047] Then, baseline drift correction is performed using a 5-level decomposition based on the db4 wavelet basis, breaking the signal down into 5 levels of varying coarseness. The low-frequency trend term at the lowest level is extracted, representing the overall drift trend of the signal. Baseline shift is eliminated using algebraic subtraction. Finally, a low-pass digital filter with a cutoff frequency of 500 MHz is used to eliminate environmental noise and high-frequency interference from the instrument.

[0048] Example 3

[0049] Based on Example 2, step 3 specifically involves: performing S-transform time-domain analysis on the preprocessed signal with a Gaussian window time-width coefficient of 0.8; locating the transient energy peak region caused by stress wave reflection-unloading coupling; then identifying the peaks and troughs of the particle velocity curve using the second-order difference method, and eliminating spurious peaks with adjacent extreme value intervals less than 0.8 μs; taking the first main peak as the starting point, selecting the time interval from 20% to 30% of the velocity decay to the peak value as the feature analysis window, and statistically analyzing the probability distribution of the amplitude difference between consecutive extreme points within this time interval.

[0050] Specifically, the peaks and troughs of the particle velocity curve are identified using the second-order difference method as follows:

[0051] First, the first-order difference is calculated on the denoised velocity curve to obtain the rate of change of the particle velocity signal; then, the first-order difference data is subjected to another difference operation to obtain the second-order difference, which reflects the curvature change of the particle velocity signal, as shown in formula (1):

[0052] (1)

[0053] In the formula, It is the particle velocity signal at point The value at the position; h is the step size.

[0054] By calculating the second-order difference, the concavity and convexity of a signal can be determined, thereby identifying peaks and troughs. Based on the sign change of the second-order difference: when it changes from positive to negative, it indicates that the signal has changed from rising to falling, corresponding to a local maximum (peak); conversely, when the second-order difference changes from negative to positive, it indicates that the signal has changed from falling to rising, corresponding to a local minimum (trough).

[0055] Secondly, after identifying the peaks and troughs, it is necessary to define a time interval that reflects the key dynamic response characteristics. The first peak in the signal is taken as the starting point, and the effective analysis interval is defined as 0.05 to 10 times the time it takes for the velocity in the curve signal to reach the peak value of the first peak. This analysis interval is crucial for accurately assessing spallation damage. The back-side particle velocities differ significantly under different impact parameters for specimens of different thicknesses. Analyzing a single set of data would significantly increase the computational load, and the constructed discrimination method would lack universality. Feasibility studies show that the falling edge of the first main peak of the back-side particle velocity curve differs significantly between undamaged and damaged states. Therefore, this invention uses the first peak in the signal as the starting point. Furthermore, the time it takes for the stress wave to complete a full propagation coupling is inconsistent in specimens of different thicknesses, requiring different cutoff times for different specimen thicknesses, which severely affects computational efficiency and the universality of the discrimination method. Therefore, this invention uses the moment when the velocity in the curve signal reaches a specific proportion of the first peak value as the cutoff time.

[0056] Within this time interval, the second-order finite difference method is used to identify all peaks and troughs in the particle velocity curve. The amplitude difference between adjacent extreme points is calculated, i.e.

[0057] (2)

[0058] In the formula, and These represent adjacent extreme points, Indicates the amplitude difference. .

[0059] The probability distribution of the amplitude difference between consecutive extreme points within the time interval is statistically analyzed using an adaptive binning method.

[0060] Based on adaptive binning, this method statistically analyzes the probability distribution of amplitude differences between consecutive extreme points within a selected time interval. Adaptive binning dynamically adjusts the width of each bin according to the data distribution characteristics, resulting in a more uniform number of data points within each bin. Compared to fixed-width binning, this method more accurately reflects the actual data distribution, especially when the data is unevenly distributed or contains outliers.

[0061] Example 4

[0062] Based on Example 3, step 4 specifically includes:

[0063] The peak entropy value is calculated using the Shannon entropy formula, reflecting the degree of disorder in the extreme value changes of the particle velocity signal, as shown below:

[0064] (3)

[0065] In the formula, The probability of the xi-th amplitude difference is used to calculate the "peak entropy," which serves as a quantitative characteristic indicator of spallation damage. A higher peak entropy value indicates more irregular extreme value changes in the signal and a more severe degree of damage.

[0066] Example 5

[0067] Based on Example 4, peak entropy The core feature is the peak entropy, and other features include time-series statistics such as extreme point density and amplitude difference kurtosis. Signal processing is completed after a single impact, and a damage probability score is output. When the peak entropy is "0", it means that the signal is relatively smooth within the selected time interval, and the corresponding specimen is in an undamaged state; when the peak entropy is greater than "0", it means that the signal is relatively chaotic within the selected time interval, and the corresponding specimen is in a damaged state.

[0068] Example 6

[0069] This embodiment uses a dynamic detection method for weak bonding defects at CFRP interfaces based on "peak entropy," specifically as follows:

[0070] S1. The laser shock uses a LABER-H50 adjustable pulse width laser with an output wavelength of 1053nm, a pulse width adjustment range of 10-300ns, an energy output range of 1-50J, and an adjustable spot diameter of 2-15mm. The dynamic signal acquisition uses a photon Doppler velocimeter (PDV) system, which includes a 1550nm continuous laser source, a 4GHz bandwidth oscilloscope (sampling rate 20GS / s), and a probe distance of 2-5mm from the back of the sample. The constraint layer system consists of a deionized water constraint layer (thickness 1-2mm) and an absorption layer of black PVC tape. The three-dimensional positioning platform has a repeatability accuracy of ±0.01mm and a load-bearing capacity of 15kg. The CT verification uses a Diondo D2 high-resolution CT system with scanning parameters of X-ray source voltage 90kV, current 90μA, and spatial resolution of 0.02mm.

[0071] S2, PDV signal acquisition and downsampling: single laser shock parameter settings, matching laser energy (1-4J), pulse width (20-100ns), and spot diameter (5mm) according to specimen thickness (1.5mm); dynamic response signal downsampling, reducing the original sampling rate from 20GS / s to 1GS / s to ensure signal bandwidth ≤500MHz (meeting the Nyquist criterion); then baseline drift correction is performed using a db4 wavelet basis for 5-level decomposition, extracting the low-frequency approximation coefficients of the 5th level as the trend term, and eliminating baseline shift through algebraic subtraction; finally, high-frequency noise suppression is achieved by designing an FIR low-pass filter (cutoff frequency 500MHz, transition bandwidth 100MHz) to eliminate high-frequency noise interference.

[0072] S3. Perform S-transform on the denoised signal (Gaussian window time width coefficient 0.8) to extract the time-domain energy matrix and locate the transient energy peak region of stress wave reflection-unloading coupling (energy threshold set to 98% of the global peak value); screen extreme points, identify wave peaks and troughs based on the second-order difference method, and remove pseudo-peaks with adjacent extreme value intervals <0.8μs; define the feature window starting from the first main wave peak and intercept the time interval from the velocity decay to 20%-30% of the peak value.

[0073] S4, for the filtered extreme point sequence Calculate the difference in amplitude between adjacent extreme values ​​using the formula:

[0074] (2)

[0075] In the formula, and These represent adjacent extreme points, Indicates the amplitude difference. .

[0076] The system dynamically divides the data into 10-15 intervals based on the amplitude difference distribution to ensure that the data volume of each bin is uniform; the normalized probability density distribution is calculated using kernel density estimation (bandwidth 0.05).

[0077] S5, calculate the Shannon entropy of the amplitude difference sequence between extreme points to generate peak entropy features, specifically:

[0078] (3)

[0079] In the formula, Indicates peak entropy characteristics, Representing the The probability of an amplitude difference. Normal specimen. ≈0 (single-peak attenuation, concentrated amplitude difference distribution); defective specimen >0 (multi-peak oscillation, amplitude difference randomly distributed).

[0080] Then input the feature peak entropy. Extreme point density (number of extreme points per unit time), amplitude difference kurtosis; signal processing is completed after a single impact, and a damage probability score (0-1) is output.

[0081] The peak entropy exhibits excellent stability under different feature window ratios, and its sensitivity to spallation damage remains consistent within a reasonable range of laser parameters. For specimens of different thicknesses, the false positive rate does not significantly increase when using the same threshold conditions.

[0082] Example 7

[0083] This embodiment uses a dynamic detection method for weak bonding defects at CFRP interfaces based on "peak entropy," employing the parameters provided in Embodiment 6, specifically:

[0084] The dynamic response signal of PDV on the back surface of the laser-shocked composite material was collected, and the free particle velocity on the back surface of the specimen under different impact conditions was extracted.

[0085] like Figure 2 and Figure 3 The figures shown are schematic diagrams for a specimen thickness of 1.5 mm and a pulse width of 50 ns, and a specimen thickness of 3 mm and a pulse width of 30 ns, respectively. Figure 2 and Figure 3 It is evident that the timing of the appearance of the particle velocity signal characteristics on the back side varies significantly under different specimen thicknesses or impact conditions. To ensure good comparability of the experimental data and to capture signal characteristics closely related to the damage mechanism, the effective analysis interval is defined as 0.05 to 10 times the time taken for the velocity in the curve signal to reach the first peak value. The first peak in the signal is determined as the starting time, and the time corresponding to the particle velocity decaying to a specific ratio (0.1-0.5) of the first peak value is defined as the ending time.

[0086] Taking a laser pulse width of 50 ns and a specimen thickness of 1.5 mm, and a laser pulse width of 30 ns and a specimen thickness of 3 mm as examples, the influence of different laser energies on peak entropy is calculated, and the detailed range of values ​​for the ratio is verified using CT data. The influence of the ratio on peak entropy is as follows: Figure 4 Show, Figure 4 The influence trend of peak entropy on the specimen under different impact parameters in both undamaged and damaged states is shown. As the laser energy increases, the peak entropy undergoes a step change. When the peak entropy of the undamaged specimen is "0", the signal disorder is low, and the specimen is in an undamaged state. When the peak entropy is greater than "0", the signal disorder increases, and the specimen is in a damaged state.

[0087] This invention provides a dynamic detection method for weak bonding defects at CFRP interfaces based on "peak entropy." It acquires dynamic response signals through a single laser impact, employs S-transform time-frequency analysis and Shannon entropy feature extraction, improving detection efficiency to 5 seconds per component. Compared to the "low-high-low" three-impact and "low-high" two-impact methods, this invention optimizes single-impact detection, improving efficiency and accuracy, and is suitable for rapid quality inspection and monitoring of aerospace CFRP components.

Claims

1. A dynamic detection method for weak bonding defects at CFRP interfaces based on "peak entropy", characterized in that, Specifically, the following steps are included: Step 1: Collect particle velocity signals on the back of the non-destructive specimen under laser shock using a photon Doppler velocimetry system; Step 2: Preprocess the particle velocity signal; Step 3: Perform S-transform time-frequency analysis and extreme value analysis on the preprocessed particle velocity signal to calculate the amplitude difference between extreme points; Step 3 specifically involves: performing S-transform time-domain analysis on the preprocessed signal to locate the transient energy peak region caused by stress wave reflection-unloading coupling; then identifying the peaks and troughs of the particle velocity curve using the second-order difference method, and eliminating spurious peaks with adjacent extreme value intervals less than 0.8 μs; taking the first main peak as the starting point, selecting the time interval from 20% to 30% of the velocity decay to the peak value as the feature analysis window, and statistically analyzing the probability distribution of the amplitude difference between consecutive extreme points within this time interval; Step 3, which describes identifying the peaks and troughs of the particle velocity curve using the second-order difference method, specifically involves: First, the first-order difference is calculated on the denoised velocity curve to obtain the rate of change of the particle velocity signal; then, the first-order difference data is subjected to another difference operation to obtain the second-order difference, which reflects the curvature change of the particle velocity signal, as shown in formula (1): （1） In the formula, It is the particle velocity signal at point The value at the position; h is the step size; The calculation of the amplitude difference between extreme points in step 3 specifically involves: （2） In the formula, and These represent adjacent extreme points, Indicates the amplitude difference. ; The probability distribution of amplitude differences between consecutive extreme points within the time interval in step 3 is statistically analyzed using an adaptive binning method. Step 4: Calculate the Shannon entropy of the amplitude difference sequence between extreme points to generate peak entropy features; Step 5: Extract peak entropy features to determine the damage state of the non-destructive specimen.

2. The dynamic detection method for weak bonding defects at CFRP interfaces based on "peak entropy" according to claim 1, characterized in that, The non-destructive test specimens mentioned in step 1 are made of carbon fiber reinforced composite materials.

3. The dynamic detection method for weak bonding defects at CFRP interfaces based on "peak entropy" according to claim 1, characterized in that, Step 2 specifically involves: first, downsampling to reduce the original signal sampling rate from 20 GS / s to 1 GS / s; then, baseline drift correction is performed, low-frequency trend terms are extracted using wavelet decomposition, and signal baseline offset is eliminated by algebraic subtraction; finally, a low-pass digital filter with a cutoff frequency of 500 MHz is used to eliminate environmental noise and high-frequency interference from the instrument.

4. The dynamic detection method for weak bonding defects at CFRP interfaces based on "peak entropy" according to claim 1, characterized in that, Step 4 specifically involves: The peak entropy value is calculated using the Shannon entropy formula, reflecting the degree of disorder in the extreme value changes of the particle velocity signal, as shown below: （3） In the formula, Representing the The probability of a difference in amplitude.

5. The dynamic detection method for weak bonding defects at CFRP interfaces based on "peak entropy" according to claim 1, characterized in that, Step 5 specifically involves: When peak entropy When a step occurs, if the peak entropy of the undamaged specimen is "0", the signal disorder is low and the specimen is in an undamaged state; if the peak entropy is greater than "0", the signal disorder increases and the specimen is in a damaged state.

Images