Thick plate structure grain size quantification method based on wake waves

Abstract

The invention provides a wake wave-based thick plate structure grain size quantification method, and belongs to the field of thick plate structure grain size quantification, and the method comprises the steps: S1, carrying out the pulse signal excitation of a thick plate structure, and collecting a wake wave signal; s2, obtaining a preliminary noise reduction signal based on the collected wake wave signal; s3, variational mode decomposition is carried out on the preliminary noise reduction signal, and a limited bandwidth signal with a specific frequency is extracted; s4, fitting the attenuation coefficient by using the average power envelope of the limited bandwidth signal; and S5, obtaining a grain size quantification result considering uncertainty by using a logarithmic linear quantification model. According to the method, accurate and rapid grain size quantification can be carried out on a thick plate structure with a relatively large frequency thickness product by utilizing the characteristics of wake waves; the overall energy attenuation rule of the signal related to the grain size is effectively described through the average power envelope, the obtained wake wave attenuation coefficient can be used for accurately and reliably quantifying the grain size of the thick plate structure, and the accuracy of the prediction result is ensured.

Description

Technical Field

[0001] The invention belongs to the technical field of thick plate structure grain size quantification, and in particular relates to a thick plate structure grain size quantification method based on wake waves. Background Art

[0002] Ensuring that the mechanical properties of the structure meet the service requirements is one of the core guarantees for the safety of key components in service. As an important microstructural parameter that affects the tensile strength and fatigue resistance of the material, accurate quantification of grain size is an important basis for improving the quality of key components and reducing development costs. Traditional grain size assessment mainly relies on time-consuming and costly destructive detection methods, such as metallography and electron backscatter diffraction methods, which are difficult to meet the needs of actual engineering applications. In contrast, ultrasonic nondestructive testing technology has become a research hotspot in the field of grain size quantification due to its high efficiency and low cost. When ultrasonic waves propagate in polycrystalline materials, they will interact with grain boundaries and cause scattering, which will cause the energy of ultrasonic waves to gradually attenuate and the signal characteristics to change. The signal characteristics of ultrasonic waves, such as attenuation, backscattered signals and wave velocity, have been shown to be correlated with grain size. Among them, attenuation is usually the preferred signal feature in industrial practice due to its intuitive principle and simple evaluation.

[0003] According to the thickness of the target structure and different combinations of ultrasonic frequencies, that is, different frequency-thickness products, ultrasonic waves will exhibit multiple propagation modes represented by body waves and guided waves. At present, for different types of products and different industrial detection needs, grain size quantification methods based on ultrasonic body waves and guided waves have been applied in specific industrial scenarios. Ultrasonic body waves are usually used at high frequency and thickness products, propagating inside the volume of the material with high spatial resolution. However, the traditional body wave method has the following limitations: first, when the frequency-thickness product is insufficient, the complexity of signal analysis is high; second, when the frequency is too high, the difficulty of quantifying grain size will increase; third, body waves can only analyze local grain information at a single measurement position, and dense sampling is required for large-scale structural evaluation, and the detection efficiency is low. Guided waves propagate along thin plate structures with low frequency and thickness products, and can efficiently detect structures in a large range, but when the frequency-thickness product is high, there will be problems such as complex propagation modes and difficulty in analysis. In summary, for thick plate structures commonly seen in engineering, it is challenging to determine the appropriate ultrasonic body wave or guided wave method for grain size evaluation.

[0004] When the frequency-thickness product in the plate structure exceeds 15MHz·mm, a wake wave will be formed in the structure, manifested as a series of trailing pulses in the ultrasonic signal. The characteristics of the wake wave are between those of the body wave and the guided wave. It has a short wavelength and long-distance propagation characteristics similar to the guided wave. It achieves an effective balance between high resolution and detection efficiency, and shows good engineering application potential in the detection of thick plate structures. However, the complex propagation mechanism of the wake wave poses a challenge to the applicability of existing signal processing methods. At present, there is no general method for quantifying grain size using the wake wave. In response to the engineering needs of quantifying grain size in thick plate materials, the study of ultrasonic detection methods based on the attenuation characteristics of the wake wave can break through the limitations of traditional body wave and guided wave methods in the detection of thick plate structures, and achieve efficient and accurate grain size quantification, which has important application value and research significance. Summary of the invention

[0005] In view of the defects in the above-mentioned prior art, the present invention proposes a grain size quantification method for thick plate structures based on wake waves. First, an ultrasonic quantification system for the grain size of thick plate structures is built, and an ultrasonic transducer is used to pulse excite the thick plate structure to be detected and collect the wake wave signal; according to the characteristics of the wake wave signal, a two-step signal noise reduction method is adopted, including cross-correlation operation with the excitation pulse and filtering the signal using the variational mode decomposition method to obtain a time-domain finite bandwidth signal; then, the attenuation law of the finite bandwidth signal after noise reduction is analyzed based on the average power envelope, so as to evaluate the attenuation coefficient of the wake wave in the thick plate structure; finally, according to the logarithmic linear quantification model between the attenuation coefficient and the grain size, the uncertainty quantification result of the grain size is obtained. Compared with the existing grain size quantification methods based on ultrasonic body waves and guided waves, the present invention is more suitable for the evaluation of thick plate structures, has higher detection efficiency, lower analysis complexity and is easy to use, and the obtained grain quantification results are highly accurate and robust.

[0006] Specifically, the present invention provides a method for quantifying grain size of thick plate structure based on wake wave, which comprises the following steps:

[0007] S1. Build an ultrasonic quantification system for grain size of thick plate structure, perform pulse signal excitation on the thick plate structure and collect wake wave signals;

[0008] S2. Perform cross-correlation operation on the collected tail wave signal f(τ) and the excitation pulse g(τ) to obtain a preliminary noise reduction signal. The calculation formula is as follows:

[0009]

[0010] Among them, τ is the time variable, t is the time shift variable;

[0011] S3, perform variational mode decomposition on the preliminary denoised signal R(t), and decompose the signal into k modes with different center frequencies ω i The bandwidth-limited intrinsic mode function s i (t), the initial noise reduction signal R(t) and the limited bandwidth intrinsic mode function s i (t) have the following relationship:

[0012]

[0013] Among them, s r (t) is the residual signal. According to the center frequency ω0 of the excitation pulse, the center frequency ω is selected from the k finite bandwidth intrinsic mode functions. i The finite bandwidth intrinsic mode function s with the smallest deviation from ω0 i (t) as a bandwidth-limited signal I(t);

[0014] S4, using the average power envelope of the finite bandwidth signal to fit the attenuation coefficient, this step includes the following sub-steps:

[0015] S41. For different time-shift variables t, the average power of the bandwidth-limited signal I(t) in the time window [t, t+T] with a width of T is calculated to obtain the average power envelope curve of the bandwidth-limited signal. The calculation formula is as follows:

[0016]

[0017] Where u is the time variable in the time window [t, t+T];

[0018] S42, based on the obtained average power envelope curve, using the e exponential decay model for fitting to obtain the attenuation coefficient α;

[0019] The calculation formula of the attenuation coefficient α is as follows:

[0020] E(t)=E0e (-αt) (4)

[0021] Among them, E0 is the fitting parameter representing the initial power; the starting time of the data used for fitting is the time of the first peak of the average power envelope; the end time is the first inflection point of the average power envelope, that is, the time when the slope is 0 for the first time; if the average power envelope continues to decay within the entire recorded time range, all data after the maximum average power are used for fitting;

[0022] S5, based on the wake wave attenuation coefficient obtained in step S4, using a logarithmic linear quantization model to obtain a grain size quantification result taking uncertainty into account, this step includes the following sub-steps:

[0023] S51, constructing a logarithmic linear quantization model, and obtaining a logarithmic quantization result of the grain size according to the logarithmic linear quantization model;

[0024] S52, calculating the probability density function of the logarithmic grain size, using the logarithmic linear quantization model to obtain multiple groups of logarithmic grain sizes, and using the normal distribution to fit the probability density function of the logarithmic grain size:

[0025]

[0026] Among them, μ and σ 2 are the mean and variance of the normal distribution respectively. According to the probability density function, the probability distribution of the predicted grain size considering the uncertainty is obtained. The mean of the probability distribution of the grain size is the predicted value of the grain size.

[0027] Preferably, the log-linear quantization model in step S51 is as follows:

[0028] ln d=m1+m2 lnα (5)

[0029] Among them, d is the grain size quantification result, m1 and m2 are the first fitting parameter and the second fitting parameter of the log-linear quantization model obtained based on the wake wave experimental evaluation of the same material, and obey the multivariate normal distribution MVN(M, ∑), where M is the mean vector and ∑ is the covariance matrix.

[0030] Preferably, in step S52, the uncertainty of the model parameters is taken into account, the first fitting parameter m1 and the second fitting parameter m2 are sampled according to the multivariate normal distribution MVN(M, ∑), and based on the solved attenuation coefficient α, the logarithmic grain size is obtained by using a log-linear quantization model using different combinations of the sampled first fitting parameters m1 and the second fitting parameters m2.

[0031] Preferably, in step S1, using a pulse signal to excite the thick plate structure to be measured and collecting the wake wave signal specifically includes the following steps:

[0032] S11, arranging an ultrasonic transducer on the surface of a target area of ​​the thick plate structure to be measured;

[0033] S12, controlling the output end of the pulse generator to generate a narrow-band sinusoidal wave pulse as an input excitation signal for the ultrasonic transducer, and generating a synchronization signal from the synchronization output end and connecting the synchronization signal to the synchronization input end of the oscilloscope;

[0034] S13, the center frequency of the excitation pulse is determined according to the thickness of the thick plate structure to be measured, the wave velocity and the estimated grain size range, so that the frequency-thickness product is greater than 15MHz·mm;

[0035] S14, using an oscilloscope to synchronously collect the tail wave signal received by the ultrasonic transducer according to the synchronization signal of the synchronization input terminal as a trigger signal, and transmit it to the upper controller.

[0036] Preferably, the tail wave signal used for cross-correlation calculation in step S2 is the signal after the initial crosstalk part of the signal is removed.

[0037] Preferably, the constrained variational problem to be constructed in the variational modal decomposition in step S3 is:

[0038]

[0039] Where δ(t) is the Dirac delta function, j is an imaginary unit, i = 1, ..., k, and by introducing the quadratic penalty term β and the Lagrange multiplier λ, the constrained problem is transformed into the following unconstrained problem:

[0040]

[0041] Among them, L({s i},{ω i},λ) is the augmented Lagrangian function.

[0042] The above optimization problem is solved iteratively to obtain k finite bandwidth intrinsic mode functions s i (t).

[0043] Preferably, in step S4, different times t for evaluating the average power are selected at a certain time interval Δt.

[0044] Preferably, the time window width T for evaluating the average power in step S4 includes a plurality of reflected wave packets.

[0045] Preferably, in step S4, the curve is averaged and smoothed, so as to suppress local noise in the average power envelope curve.

[0046] Preferably, in step S5, the first fitting parameter and the second fitting parameter of the model and their respective distributions are determined using a plurality of thick plate specimens having the same thickness and material as the thick plate structure to be measured but with different grain sizes.

[0047] Compared with the prior art, the beneficial technical effects of the present invention are:

[0048] (1) The wake wave-based grain size quantification method for thick plate structures proposed in the present invention utilizes the short wavelength and long-distance propagation characteristics of wake waves to overcome the detection limitations of traditional body wave and waveguide technologies for thick plate structures, and is conducive to accurate and rapid grain size quantification of thick plate structures with large frequency-thickness products. The overall energy attenuation law related to the signal and the grain size is effectively described by the average power envelope, and the obtained wake wave attenuation coefficient can be used to accurately and reliably quantify the grain size of the thick plate structure.

[0049] (2) The two-step signal denoising method proposed in the method of the present invention simplifies the attenuation characteristic analysis process of complex wake wave signals, wherein the cross-correlation operation with the excitation pulse effectively suppresses the incoherent noise, and the variational mode decomposition decomposes the signal into a simple finite bandwidth signal while retaining the key attenuation characteristics; by extracting finite bandwidth signals with similar frequency components for subsequent attenuation analysis, the quantization error caused by the frequency difference of different specimens can be reduced, thereby improving the accuracy and robustness of grain size quantization. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Other features, objects and advantages of the present application will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings.

[0051] Figure 1 It is a flow chart of the lossless quantification method of thick plate structure grain size based on wake wave of the present invention;

[0052] Figure 2 It is a schematic diagram of an ultrasonic quantification system and a process flow of a thick plate structure grain size in an embodiment of the present invention;

[0053] Figure 3 is a schematic diagram of the waveform and power spectrum density of the excitation signal in an embodiment of the present invention;

[0054] Figure 4 is a waveform and power spectrum density diagram of a wake wave signal in a thick plate structure collected in an embodiment of the present invention;

[0055] Figure 5 1 is a schematic diagram of the waveform and power spectrum density of the cross-correlation operation result of the tail wave signal in an embodiment of the present invention;

[0056] Figure 6 is a waveform and power spectrum density diagram of a variational modal decomposition processing result of a noise reduction signal in an embodiment of the present invention;

[0057] Figure 7 is a schematic diagram of average power envelope fitting of a bandwidth-limited signal in an embodiment of the present invention;

[0058] Figure 8It is the probability density function result of the quantification of the grain size of the thick plate structure obtained in the embodiment of the present invention. DETAILED DESCRIPTION

[0059] The present application will be further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the relevant invention, rather than to limit the invention. It should also be noted that, for ease of description, only the parts related to the relevant invention are shown in the accompanying drawings.

[0060] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0061] The present invention proposes a method for quantifying the grain size of thick plate structures based on wake waves. Figure 1 As shown, the method comprises the following steps:

[0062] S1. Build an ultrasonic quantification system for grain size of thick plate structure, use pulse signal to excite and collect wake wave signal.

[0063] like Figure 2 As shown, the ultrasonic quantification system of grain size of thick plate structure includes a pulse generator, an ultrasonic transducer, an oscilloscope and a host controller. The pulse generator generates a pulse excitation electrical signal, and the output end is transmitted to the excitation end of the ultrasonic transducer through a wire, generating ultrasonic waves in the thick plate structure to be measured; the receiving end of the ultrasonic transducer collects the ultrasonic waves in the structure and forms an electrical signal, which is connected to the input end of the oscilloscope through a wire and collected; the synchronous output end of the pulse generator is connected to the synchronous input end of the oscilloscope to ensure synchronous operation between different devices; the host controller is used to adjust the output signal of the pulse generator and record the collected signal of the oscilloscope for analysis.

[0064] Collecting ultrasonic wake wave signals specifically includes the following sub-steps:

[0065] S11. Arrange an ultrasonic transducer on the surface of a target area of ​​the thick plate structure to be measured.

[0066] S12, controlling the output end of the pulse generator to generate a narrow-band sinusoidal wave pulse as an input excitation signal for the ultrasonic transducer, and at the same time generating a synchronization signal from the synchronization output end and connecting it to the synchronization input end of the oscilloscope.

[0067] S13. The center frequency of the excitation pulse is determined according to the thickness of the thick plate structure to be measured, the wave velocity and the estimated grain size range, so that the frequency-thickness product is greater than 15 MHz·mm, so as to ensure the formation of a wake wave in the structure and improve the accuracy of grain size measurement.

[0068] S14, using an oscilloscope to synchronously collect the tail wave signal received by the ultrasonic transducer according to the synchronization signal of the synchronization input terminal as a trigger signal, and transmit it to the upper controller.

[0069] S2. After removing the initial crosstalk part of the signal, the collected tail wave signal f(τ) and the excitation pulse g(τ) are cross-correlated to obtain a preliminary noise reduction signal. The calculation formula is as follows:

[0070]

[0071] Among them, τ is the time variable and t is the time shift variable, both of which are variables representing time.

[0072] S3, performing variational mode decomposition on the preliminary noise reduction signal R(t) to extract a limited bandwidth signal with a specific frequency, which step includes the following sub-steps:

[0073] S31. Construct the following constrained variational problem: decompose the signal into k components with different center frequencies ω. i The bandwidth-limited intrinsic mode function s i (t), as follows:

[0074]

[0075] Where δ(t) is the Dirac delta function, j is the imaginary unit, i = 1, ..., k, and by introducing the quadratic penalty term β and the Lagrange multiplier λ, the constrained problem is transformed into the following unconstrained problem:

[0076]

[0077] The above optimization problem is solved iteratively to obtain k finite bandwidth intrinsic mode functions s i (t), and the following relationship exists:

[0078]

[0079] Among them, s r (t) is the residual signal.

[0080] S32, according to the center frequency ω0 of the excitation pulse, select the center frequency ω from the k finite bandwidth natural mode functions i The bandwidth intrinsic mode function s of the finite bandwidth signal with the smallest deviation from ω0 i (t) as I(t).

[0081] S4, using the average power envelope of the finite bandwidth signal to fit the attenuation coefficient, this step includes the following sub-steps:

[0082] S41. At different times t, the average power of the bandwidth-limited signal I(t) in the time window [t, t+T] with a width of T is calculated to obtain an average power envelope curve of the bandwidth-limited signal. The calculation formula is as follows:

[0083]

[0084] Where y is the time variable in the time window [t, t+T]. The width of the time window T should include multiple reflected wave packets to reduce interference. In order to suppress local noise in the average power envelope curve, the curve can be averaged and smoothed.

[0085] S42. Based on the obtained average power envelope curve, an exponential decay model is used for fitting to obtain an attenuation coefficient α. The calculation formula is as follows:

[0086] E(t)=E0e (-αt) (4)

[0087] Among them, E0 is the fitting parameter representing the initial power. In order to ensure a single decay mode and improve the quantification accuracy of the decay coefficient, the start time for fitting the data is defined as the time of the first peak of the average power envelope; the end time is defined as the first inflection point of the average power envelope, that is, the time when the slope is 0 for the first time. If the average power envelope continues to decay within the entire recorded time range, all data after the maximum average power are used for fitting.

[0088] S5, based on the wake wave attenuation coefficient obtained in step S4, using a logarithmic linear quantization model to obtain a grain size quantification result taking uncertainty into account, this step includes the following sub-steps:

[0089] S51. Obtain the mean of the logarithmic quantization results of the grain size according to the logarithmic linear quantization model. The logarithmic linear quantization model is as follows:

[0090] ln d=m1+m2 lnα (5)

[0091] Among them, d is the grain size quantification result, m1 and m2 are the first fitting parameter and the second fitting parameter of the log-linear quantization model obtained based on the wake wave experimental evaluation of the same material, and obey the multivariate normal distribution MVN(M, ∑), where M is the mean vector and ∑ is the covariance matrix.

[0092] S52. Considering the uncertainty of the model parameters, the first fitting parameter m1 and the second fitting parameter m2 are sampled according to the multivariate normal distribution MVN(M, ∑). According to the given attenuation coefficient α, the combination of the different first fitting parameters m1 and the second fitting parameters m2 obtained by sampling is used to evaluate the logarithmic grain size based on formula (7), and then the probability density function of the logarithmic grain size is fitted using the normal distribution:

[0093]

[0094] Among them, μ and σ 2 are the mean and variance of the normal distribution, respectively. According to the probability density function, the probability distribution of the grain size quantification result considering the uncertainty can be obtained. The mean of the probability distribution of the grain size is the predicted value of the grain size. At the same time, the prediction confidence can be obtained according to the probability distribution of the grain size to evaluate the prediction accuracy. Specific embodiments

[0096] The present invention will be further described in detail below with reference to a specific case of a thick plate structure grain size quantification method based on wake waves.

[0097] S1. The thick plate structure to be tested is a nickel-based high-temperature alloy specimen with a thickness of 5 mm. An ultrasonic quantification system for the grain size of the thick plate structure is built to perform ultrasonic testing on the thick plate structure to be tested. The specific process is as follows:

[0098] S11. Use 502 glue to stick a pair of ultrasonic transducers including an exciter and a receiver to the center of the surface of the thick plate structure to be tested to form a pulse echo detection configuration.

[0099] S12, control the output end of the pulse generator (Tektronix, AFG 31022) to generate a 3.5-cycle Hanning window modulated narrow-band sine wave pulse as the ultrasonic transducer input excitation signal, and at the same time generate a synchronization signal from the synchronization output end and connect it to the synchronization input end of the oscilloscope (Tektronix, MDO3104).

[0100] S13. The center frequency of the excitation pulse is selected as 5MHz based on the thickness of the thick plate structure to be tested, the wave velocity and the estimated grain size range. The frequency-thickness product is 25MHz·mm, which can meet the conditions for forming a wake wave in the structure and is sensitive to tiny grain sizes. The time domain waveform and spectrum of the excitation pulse g(t) are as follows: Figure 3 shown.

[0101] S14, using an oscilloscope to synchronously collect the wake wave signal received by the ultrasonic transducer according to the synchronization signal of the synchronization input terminal as a trigger signal. In order to improve the signal-to-noise ratio of the signal, the collected ultrasonic signal is firstly processed by 128 times of time domain averaging in the oscilloscope, and then transmitted to the computer for subsequent processing. The waveform and power spectrum density of the wake wave signal f(t) collected in the thick plate structure to be tested are shown in Figure 4 shown.

[0102] S2. According to formula (1), the collected tail wave signal f(t) and the excitation pulse g(t) are cross-correlated, and the crosstalk signal before 1.5 μs is eliminated to obtain the preliminary noise reduction signal R(t). The result is as follows: Figure 5 As shown in the figure, the time domain waveform of the signal after preliminary noise reduction maintains the main features consistent with the original tail wave signal and retains the components related to the input pulse. In the frequency domain, the preliminary noise reduction signal retains the 3-6MHz peak with the highest energy, while other incoherent noise outside this range is relatively suppressed.

[0103] S3, such as Figure 6 As shown in the figure, according to formula (3), the initial noise reduction signal R(t) is subjected to variational mode decomposition to obtain six limited bandwidth intrinsic mode functions s i (t) (abbreviated as IMF) and the residual signal s r (t), i = 1, ..., 6. According to Figure 6 In the spectrum diagram of each finite bandwidth natural mode function shown, and the center frequency of the excitation pulse is 5 MHz, the third finite bandwidth signal I(t)=s3(t) with the smallest center frequency deviation is selected for subsequent analysis.

[0104] S4, select the time window width T = 17.0μs, which is ten times the original signal pulse packet interval. According to formula (5), with Δt = 0.02μs as the interval, the average power of the finite bandwidth signal I(t) in the time window [t, t+T] is calculated at different times t, and after 5-point average smoothing, the average power envelope curve of the finite bandwidth signal is obtained, as shown in Figure 7 The average power envelope curve from 1.5μs to 19.72μs is intercepted and fitted using the exponential decay model shown in formula (6). The fitting result is shown in Figure 7 As shown, the attenuation coefficient α is obtained as 0.34 μs -1 .

[0105] S5. In order to obtain the first fitting parameter m1 and the second fitting parameter m2 in the logarithmic linear quantization model shown in formula (7), 10 5mm thick plate specimens T1-T10 of the same material with different grain sizes were prepared, and the same ultrasonic testing steps as above were performed to obtain the corresponding attenuation coefficients. The first fitting parameter m1 and the second fitting parameter m2 of the model were estimated by maximum likelihood based on the attenuation coefficient and grain size, and the estimated values ​​m1=7.14 and m2=1.98 were obtained, and they obeyed the mean vector M=[7.14, 1.98], and the covariance matrix According to the multivariate normal distribution MVN(M,∑), the first fitting parameter m1 and the second fitting parameter m2 are subjected to 5×10 6The logarithmic grain size is evaluated based on formula (7) using the combination of different first fitting parameters m1 and second fitting parameters m2 obtained by sampling, and then the probability density function of the logarithmic grain size is fitted using the normal distribution to obtain the probability distribution of the logarithmic grain size quantification result considering the uncertainty. Figure 8 The mean and variance of the normal distribution are μ = 5.01 and σ respectively. 2 =0.15. The true grain size in the thick plate structure to be measured was determined to be 159.18 μm using the electron backscatter diffraction technique. It can be observed that the logarithmic value of the true grain size is within the 95% confidence interval of the probability density function, which illustrates the effectiveness and robustness of the log-linear model. According to the maximum value of the probability density function, the average grain size predicted by the model is 149.60 μm, which is close to the true value of 159.18 μm, with a relative error of only 6.02%. This shows that the proposed method can effectively and accurately quantify the grain size of thick plate structures.

[0106] The present invention proposes a grain size quantification method for thick plate structure based on wake wave, which builds an ultrasonic quantification system for grain size of thick plate structure, uses an ultrasonic transducer to pulse excite the thick plate structure to be detected and collects the wake wave signal; according to the characteristics of the wake wave signal, a two-step signal noise reduction method is adopted, including cross-correlation operation with the excitation pulse and filtering the signal by variational mode decomposition method to obtain a time domain finite bandwidth signal; then, the attenuation law of the finite bandwidth signal after noise reduction is analyzed based on the average power envelope, so as to evaluate the attenuation coefficient of the wake wave in the thick plate structure; finally, according to the logarithmic linear quantization model between the attenuation coefficient and the grain size, the uncertainty quantification result of the grain size is obtained. Compared with the existing grain size quantification method based on ultrasonic body wave and guided wave, the present invention is more suitable for the evaluation of thick plate structure, has higher detection efficiency, lower analysis complexity and is easy to use, and the obtained grain quantification result has high accuracy and strong robustness.

[0107] Finally, it should be noted that the above embodiments are only intended to illustrate rather than limit the technical solutions of the present invention. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that the present invention can still be modified or replaced by equivalents. Any modification or partial replacement that does not depart from the spirit and scope of the present invention should be included in the scope of the claims of the present invention.

Claims

1. A method for quantifying grain size of thick plate structure based on wake wave, characterized in that: It includes the following steps: S1. Build an ultrasonic quantification system for grain size of thick plate structure, perform pulse signal excitation on the thick plate structure and collect wake wave signals; S2. Perform cross-correlation operation on the collected tail wave signal f(τ) and the excitation pulse g(τ) to obtain a preliminary noise reduction signal. The calculation formula is as follows: Among them, R(t) is the preliminary noise reduction signal, τ is the time variable, and t is the time shift variable; S3, perform variational mode decomposition on the preliminary denoised signal R(t), and decompose the signal into k modes with different center frequencies ω i The bandwidth-limited intrinsic mode function s i (t), the initial noise reduction signal R(t) and the limited bandwidth intrinsic mode function s i (t) have the following relationship: Among them, s r (t) is the residual signal. According to the center frequency ω0 of the excitation pulse, the center frequency ω is selected from the k finite bandwidth intrinsic mode functions. i The finite bandwidth intrinsic mode function s with the smallest deviation from ω0 i (t) as a bandwidth-limited signal I(t); S4, using the average power envelope of the finite bandwidth signal to fit the attenuation coefficient, this step includes the following sub-steps: S41. For different t, the average power of the bandwidth-limited signal I(t) in the time window [t, t+T] with a width of T is calculated to obtain the average power envelope curve of the bandwidth-limited signal. The calculation formula is as follows: Where, u is the time variable in the time window [t, t+T], and E(t) is the average power envelope curve; S42, based on the obtained average power envelope curve, using the e exponential decay model for fitting to obtain the attenuation coefficient α; The calculation formula of the attenuation coefficient α is as follows: E(t)=E0e (-αt) (4) Among them, E0 is the fitting parameter representing the initial power; the starting time of the data used for fitting is the time of the first peak of the average power envelope; the end time is the first inflection point of the average power envelope, that is, the time when the slope is 0 for the first time; if the average power envelope continues to decay within the entire recorded time range, all data after the maximum average power are used for fitting; S5, based on the wake wave attenuation coefficient obtained in step S4, using a logarithmic linear quantization model to obtain a grain size quantification result taking uncertainty into account, this step includes the following sub-steps: S51, constructing a logarithmic linear quantization model, and obtaining a logarithmic quantization result lnd of the grain size according to the logarithmic linear quantization model; S52, calculating the probability density function of the logarithmic grain size, using the logarithmic linear quantization model to obtain multiple groups of logarithmic grain sizes, and using the normal distribution to fit the probability density function of the logarithmic grain size: Among them, μ and σ 2 are the mean and variance of the normal distribution respectively. According to the probability density function, the probability distribution of the predicted grain size considering the uncertainty is obtained. The mean of the probability distribution of the grain size is the predicted value of the grain size.

2. The method for quantifying grain size of thick plate structure based on wake wave according to claim 1, characterized in that: The log-linear quantization model in step S51 is as follows: lnd=m1+m2 lnα (5) Among them, d is the grain size quantification result, m1 and m2 are the first fitting parameter and the second fitting parameter of the log-linear quantization model obtained based on the wake wave experimental evaluation of the same material, and obey the multivariate normal distribution MVN(Μ,Σ), where Μ is the mean vector and Σ is the covariance matrix.

3. The method for quantifying grain size of thick plate structure based on wake wave according to claim 1, characterized in that: In step S52, considering the uncertainty of the model parameters, the first fitting parameter m1 and the second fitting parameter m2 are sampled according to the multivariate normal distribution MVN(Μ,Σ), and based on the solved attenuation coefficient α, the logarithmic grain size is obtained by using the logarithmic linear quantization model using the different combinations of the sampled first fitting parameters m1 and the second fitting parameters m2.

4. The method for quantifying grain size of thick plate structure based on wake wave according to claim 1, characterized in that: In step S1, the thick plate structure to be measured is excited by a pulse signal and the wake wave signal is collected, which specifically includes the following steps: S11, arranging an ultrasonic transducer on the surface of a target area of ​​the thick plate structure to be measured; S12, controlling the output end of the pulse generator to generate a narrow-band sinusoidal wave pulse as an input excitation signal for the ultrasonic transducer, and generating a synchronization signal from the synchronization output end and connecting the synchronization signal to the synchronization input end of the oscilloscope; S13, the center frequency of the excitation pulse is determined according to the thickness of the thick plate structure to be measured, the wave velocity and the estimated grain size range, so that the frequency-thickness product is greater than 15MHz·mm; S14, using an oscilloscope to synchronously collect the tail wave signal received by the ultrasonic transducer according to the synchronization signal of the synchronization input terminal as a trigger signal, and transmit it to the upper controller.

5. The method for quantifying grain size of thick plate structure based on wake wave according to claim 1, characterized in that: The tail wave signal used for cross-correlation calculation in step S2 is the signal after the initial crosstalk part of the signal is removed.

6. The method for quantifying grain size of thick plate structure based on wake wave according to claim 1, characterized in that: The constrained variational problem to be constructed for variational modal decomposition in step S3 is: Where δ(t) is the Dirac delta function, j is the imaginary unit, i = 1,…,k, and by introducing the quadratic penalty term β and the Lagrange multiplier λ, the constrained problem is transformed into the following unconstrained problem: Among them, L({s i },{ω i },λ) is the augmented Lagrangian function, and the optimization problem is solved iteratively to obtain k finite bandwidth intrinsic mode functions s i (t).

7. The method for quantifying grain size of thick plate structure based on wake wave according to claim 1, characterized in that: In step S4, different times t for evaluating the average power are selected at a certain time interval Δt.

8. The method for quantifying grain size of thick plate structure based on wake wave according to claim 1, characterized in that: The time window width T used for evaluating the average power in step S4 includes a plurality of reflected wave packets.

9. The method for quantifying grain size of thick plate structure based on wake wave according to claim 1, characterized in that: In step S4, the curve is averaged and smoothed, so that local noise in the average power envelope curve can be suppressed.

10. The method for quantifying grain size of thick plate structure based on wake wave according to claim 1, characterized in that: In step S5, the first fitting parameter and the second fitting parameter of the model and their respective distributions are determined using a plurality of thick plate specimens having the same thickness and material as the thick plate structure to be measured but with different grain sizes.

Images