Stress detection method and system based on laser ultrasound and support vector regression

Abstract

The invention relates to the technical field of stress detection, in particular to a stress detection method and system based on laser ultrasound and support vector regression. The stress detection method based on laser ultrasound and support vector regression comprises the following steps: acquiring ultrasonic signals of an additive manufacturing material under different stresses; based on the ultrasonic signal, determining an acoustic delay characteristic of the ultrasonic signal in a composite domain; based on the acoustic delay characteristics in the composite domain and a support vector machine regression model, constructing a mapping relation between delay and stress change; acquiring a current ultrasonic signal of the additive manufacturing material; and determining a stress value in the additive manufacturing material based on the mapping relation in combination with the current ultrasonic signal. And non-contact and high-precision nondestructive stress detection can be carried out on the additive manufacturing material in a complex environment.

Description

Technical Field

[0001] The present application relates to the field of stress detection technology, and in particular to a stress detection method and system based on laser ultrasound and support vector regression. Background Art

[0002] Laser additive manufacturing has the advantages of high design freedom, high resource utilization, and the ability to achieve rapid prototyping of complex parts, showing broad application prospects in key fields such as aerospace. However, in the additive manufacturing process, the high-intensity laser beam concentrates energy to locally melt the metal material, and the subsequent rapid cooling causes a significant temperature gradient inside the material, which in turn forms residual stress. Residual stress may cause problems such as cracks and deformation, seriously affecting the performance and life of the parts.

[0003] Traditional residual stress detection methods, such as X-ray, neutron diffraction and ultrasound, have some limitations. Although X-ray and neutron diffraction methods can detect deeper stress, their equipment costs are high, they are highly radioactive, and they are not adaptable to complex geometries and high-temperature environments. Traditional contact ultrasonic methods are also limited by the probe's working stability in high-temperature environments and the temperature resistance of the coupling agent.

[0004] Therefore, it is urgent to propose a new stress detection method to solve the above technical problems. Summary of the invention

[0005] The embodiments of the present application provide a stress detection method and system based on laser ultrasound and support vector regression, which can adapt to high temperatures and can perform non-contact, high-precision, non-destructive stress detection on additive manufacturing materials in complex environments.

[0006] In some embodiments, a stress detection method based on laser ultrasound and support vector regression is provided, comprising: Obtain ultrasonic signals of additive manufacturing materials under different stresses; Based on the ultrasonic signal, determining the acoustic delay characteristics of the ultrasonic signal in the composite domain; Based on the acoustic delay characteristics in the composite domain and the support vector machine regression model, the mapping relationship between delay and stress change is constructed; Acquire the current ultrasonic signal of the additive manufacturing material; Based on the mapping relationship and combined with the current ultrasonic signal, the stress value in the additive manufacturing material is determined.

[0007] The stress detection method based on laser ultrasound and support vector regression provided in the embodiment of the present application is adopted. By analyzing the ultrasonic signal of the additive manufacturing material under different stresses, the ultrasonic sound delay in the composite domain is determined, and then combined with the support vector regression (SVR) model, a mapping relationship between stress and ultrasonic sound delay is established, thereby realizing the prediction of unknown stress and improving the efficiency and accuracy of online stress detection in the additive manufacturing process. Long-distance, non-contact, non-destructive detection of residual stress of materials in the additive manufacturing process is realized.

[0008] Optionally, the composite domain includes a time domain, a frequency domain and a time-frequency domain; based on the ultrasonic signal, determining the acoustic delay characteristics of the ultrasonic signal in the composite domain includes: Preprocessing the ultrasonic signal to determine the time domain signal corresponding to the ultrasonic signal; In the time domain, the time domain sound delay is determined based on the cross-correlation analysis and the time domain signal; In the frequency domain, based on the time domain signal, the frequency domain sound delay is determined; In the time-frequency domain, the time-frequency domain sound delay is calculated based on the continuous wavelet transform technology and the time-domain signal; Among them, the sound delay characteristics include time domain sound delay, frequency domain sound delay and time-frequency domain sound delay.

[0009] Optionally, the ultrasonic signal includes a first ultrasonic signal of the additive manufacturing material under a first preset stress and a second ultrasonic signal under a second preset stress; the first preset stress is 0, and the second preset stress is non-0; In the time domain, the time domain sound delay is determined based on the cross-correlation analysis and the time domain signal, including: Determining a first arrival time of the first ultrasonic signal and a second arrival time of the second ultrasonic signal based on the time domain signal; Perform cross-correlation calculation based on the time domain signal to determine the cross-correlation peak value; The time domain acoustic delay is calculated based on the cross-correlation peak, the first arrival time and the second arrival time.

[0010] Optionally, the ultrasonic signal includes a first ultrasonic signal of the additive manufacturing material under a first preset stress and a second ultrasonic signal under a second preset stress; the first preset stress is 0, and the second preset stress is not 0; the first time domain signal corresponds to a first frequency domain spectrum, and the second time domain signal corresponds to a second frequency domain spectrum; In the frequency domain, based on the time domain signal, the frequency domain sound delay is determined, including: Based on the time domain signal, the frequency domain spectrum corresponding to the time domain signal is determined by using Fourier transform technology; Based on the frequency domain spectrum, a cross-correlation analysis is performed to determine the frequency domain spectrum cross-correlation analysis result; Determine a phase change value of the frequency domain spectrum under a second preset stress corresponding to the frequency domain spectrum cross-correlation analysis result; Based on the phase change value, a frequency domain sound delay curve is calculated and determined; Determine a center frequency of the second time domain signal based on a second frequency domain spectrum corresponding to the second time domain signal; The frequency domain sound delay is determined based on the frequency domain sound delay curve and the center frequency of the second time domain signal.

[0011] Optionally, the ultrasonic signal includes a first ultrasonic signal of the additive manufacturing material under a first preset stress and a second ultrasonic signal under a second preset stress; the first preset stress is 0, and the second preset stress is non-0; In the time-frequency domain, based on the continuous wavelet transform technology and time-domain signals, the time-frequency domain sound delay is calculated, including: Using continuous wavelet transform, the time domain signal is converted into a time-frequency domain signal composed of wavelet coefficients; Determine the maximum value of the wavelet coefficient corresponding to the time-frequency domain signal; Determine a reference time corresponding to the maximum value of the wavelet coefficient under the first preset stress, and a comparison time corresponding to the maximum value of the wavelet coefficient under the second preset stress; Based on the reference time and the comparison time, the time-frequency domain sound delay is calculated.

[0012] Optionally, preprocessing the ultrasonic signal to determine a time domain signal corresponding to the ultrasonic signal includes: Decomposing the ultrasonic signal based on the Symlets wavelet basis function to determine multiple ultrasonic sub-signals corresponding to the ultrasonic signal; Processing each ultrasonic sub-signal based on wavelet threshold respectively; Based on using inverse wavelet transform and multiple ultrasound sub-signals, a time domain signal is reconstructed.

[0013] Optionally, obtaining ultrasonic signals of the additive manufacturing material under different stresses includes: When the additive manufacturing material has a first preset stress, collecting a first centering signal of a first point detection source and a first same-side signal of a second point detection source; When the additive manufacturing material has a second preset stress, collecting a second centering signal of the first point detection source and a second same-side signal of the second point detection source; The first preset stress magnitude is 0, and the second preset stress magnitude is non-zero.

[0014] Optionally, based on the acoustic delay characteristics in the composite domain and the support vector machine regression model, a mapping relationship between the delay and the stress change is constructed, including: Based on the acoustic delay characteristics and random recombination strategy in the composite domain, a random recombination training set is determined; The randomly reorganized training set is input into the initial support vector machine regression model, the initial support vector machine regression model is trained, and the mapping relationship is determined.

[0015] Optionally, based on the mapping relationship and in combination with the current ultrasonic signal, determining the stress value in the additive manufacturing material includes: Acquire the current ultrasonic signal of the additive manufacturing material; Combined with the mapping relationship, the stress value corresponding to the current ultrasonic signal is determined.

[0016] In some embodiments, a stress detection system based on laser ultrasound and support vector regression is provided, comprising: A laser ultrasonic detection module is configured to collect laser ultrasonic signals of additive manufacturing materials with different stresses; A demodulation device is communicatively connected with the laser ultrasonic detection module; the demodulation device is configured to obtain ultrasonic signals of additive manufacturing materials under different stresses; based on the ultrasonic signals, the acoustic delay characteristics of the ultrasonic signals in the composite domain are determined; based on the acoustic delay characteristics in the composite domain and a support vector machine regression model, a mapping relationship between delay and stress change is constructed; the current ultrasonic signal of the additive manufacturing material is obtained; based on the mapping relationship, the stress value in the additive manufacturing material is determined in combination with the current ultrasonic signal.

[0017] It can be understood that the beneficial effects that can be produced by the stress detection system based on laser ultrasound and support vector regression provided in the embodiment of the present application can be referred to the stress detection method based on laser ultrasound and support vector regression provided in the above embodiment and any optional implementation method thereof, and the present application will not repeat them here. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solution of the present application, the drawings required for use in the embodiments are briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0019] Figure 1 A first structural block diagram of a stress detection system based on laser ultrasound and support vector regression provided in an embodiment of the present application; Figure 2 A schematic diagram of ultrasonic testing of additively manufactured materials using a laser ultrasonic testing module provided in an embodiment of the present application; Figure 3 A flowchart of a stress detection method based on laser ultrasound and support vector regression provided in an embodiment of the present application; Figure 4 A second structural block diagram of a stress detection system based on laser ultrasound and support vector regression provided in an embodiment of the present application; Figure 5 A schematic diagram of a process for detecting stress of additively manufactured materials provided in an embodiment of the present application.

[0020] Reference numerals: 1. Stress detection system based on laser ultrasound and support vector regression; 11. Laser ultrasonic detection module; 111. Laser excitation line source; 112. First point detection source; 113. Second point detection source; 12. Demodulation equipment; 121. Signal processing module; 122. Stress prediction module; 123. Industrial control computer; 13. Additive manufacturing materials. DETAILED DESCRIPTION

[0021] The technical solutions in the embodiments of the present application will be described clearly below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments of the present application, other embodiments obtained by ordinary technicians in this field without making creative work all belong to the protection scope of the present application.

[0022] In the following, the terms "first", "second", etc. are used for descriptive purposes only and are not to be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first", "second", etc. may explicitly or implicitly include one or more of the features. In the description of this application, unless otherwise specified, "plurality" means two or more.

[0023] In addition, in the present application, directional terms such as "upper", "lower", "inner" and "outer" are defined relative to the orientation of the components schematically placed in the drawings. It should be understood that these directional terms are relative concepts. They are used for relative description and clarification, and they can change accordingly according to the changes in the orientation of the components placed in the drawings.

[0024] Figure 1 This is a first structural block diagram of a stress detection system based on laser ultrasound and support vector regression provided in an embodiment of the present application.

[0025] Figure 2 Schematic diagram of ultrasonic testing of additively manufactured materials using a laser ultrasonic testing module provided in an embodiment of the present application.

[0026] Combination Figure 1 and Figure 2As shown, an embodiment of the present application provides a stress detection system 1 based on laser ultrasound and support vector regression, including: a laser ultrasonic detection module 11 and a demodulation device 12. The laser ultrasonic detection module 11 is configured to collect laser ultrasonic signals for additive manufacturing materials with different stresses. The demodulation device 12 is communicatively connected to the laser ultrasonic detection module 11. The demodulation device 12 is configured to obtain ultrasonic signals of additive manufacturing materials under different stresses, determine the stress corresponding to the ultrasonic signal based on the composite domain acoustic delay technology and the support vector machine regression technology, so as to detect the residual stress in the additive manufacturing material.

[0027] The stress detection system based on laser ultrasound and support vector regression provided in the embodiment of the present application is used to realize long-distance, non-contact, non-destructive detection of residual stress of materials in the additive manufacturing process, and the system can be applied to high temperature and complex environments.

[0028] It is understandable that the demodulation device 12 includes but is not limited to the industrial control computer 123, and only needs to be able to achieve the above functions, and this application does not impose any restrictions on this.

[0029] Based on the acoustic elastic effect of solids, when stress exists inside the material, the main effect of stress on ultrasonic propagation is the change in propagation velocity. Generally speaking, the relative velocity change is proportional to the stress change. Therefore, the relative change in ultrasonic velocity is proportional to the uniaxial stress. The following relationship exists: ; in, and are the ultrasonic sound speeds in the stressed and unstressed states, respectively. and They represent the propagation direction and polarization direction of the ultrasonic wave respectively. is the acoustic elastic coefficient of the material. When the propagation distance of ultrasound remains unchanged, the change in sound velocity can be measured according to the difference in ultrasound propagation time, that is, the delay in the arrival time of ultrasound compared to the stress-free material.

[0030] Therefore, by comparing the arrival time of ultrasonic sound waves under stress and stress-free states, the delay in the arrival time of ultrasonic waves due to the presence of stress can be determined, thereby constructing a mapping relationship between stress and ultrasonic sound delay, so as to predict the residual stress in the material based on the ultrasonic signal.

[0031] Figure 3 A flowchart of a stress detection method based on laser ultrasound and support vector regression provided in an embodiment of the present application.

[0032] Combination Figure 1The stress detection system based on laser ultrasound and support vector regression is shown. The embodiment of the present application also provides a stress detection method based on laser ultrasound and support vector regression, such as Figure 3 As shown, the stress detection method based on laser ultrasound and support vector regression includes steps S1 to S5, which are specifically as follows: Step S1, obtaining ultrasonic signals of additive manufacturing materials under different stresses.

[0033] In this step, the ultrasonic signal is the first wave signal of the ultrasound.

[0034] Optionally, the ultrasonic signal includes a first ultrasonic signal of the additive manufacturing material under a first preset stress and a second ultrasonic signal under a second preset stress, the first preset stress is 0, and the second preset stress is non-zero.

[0035] Combination Figure 2 As shown, a laser excitation line source 111 is used to output laser ultrasound to the additive manufacturing material 13, a second point detection source 113 on the same side as the laser excitation line source 111 acquires an ultrasound ipsilateral signal, and a first point detection source 112 on the opposite side of the laser excitation line source 111 acquires an ultrasound centroid signal, so as to determine the acoustic delay characteristics of the composite domain based on the ipsilateral signal or the centroid signal.

[0036] Exemplarily, the ultrasonic signal includes a same-side signal or a centroid signal, the same-side signal includes a first same-side signal and a second same-side signal, and the centroid signal includes a first centroid signal and a second centroid signal. The same-side signal is an ultrasonic signal detected by the point detection source when the laser excitation source and the point detection source are located on the same side of the additive manufacturing material. The centroid signal is an ultrasonic signal detected by the point detection source when the laser excitation source and the point detection source are located on opposite sides of the additive manufacturing material. The first same-side signal and the first centroid signal correspond to the additive manufacturing material having a first preset stress, and the second same-side signal and the second centroid signal correspond to the additive manufacturing material having a second preset stress.

[0037] Optionally, the first preset stress magnitude is 0, and the second preset stress magnitude is non-0. Step S1 includes step S11 and step S21, which are specifically as follows: Step S11, when the additive manufacturing material has a first preset stress, collecting a first centering signal of a first point detection source and a first same-side signal of a second point detection source.

[0038] Step S12, when the additive manufacturing material has a second preset stress, collecting a second centering signal of the first point detection source and a second same-side signal of the second point detection source.

[0039] In this embodiment, ultrasonic signals of additively manufactured materials under different stresses are obtained to facilitate determination of ultrasonic acoustic delay within a composite domain.

[0040] Exemplarily, the positions of the first point detection source 112 and the second point detection source 113 are as follows: Figure 2 shown.

[0041] It is worth noting that in the description of the following embodiments, the ultrasonic signal is an ipsilateral signal or an aligning-heart signal, that is, in the following embodiments, the processing of the ultrasonic signal only represents the processing of the first ipsilateral signal and the second ipsilateral signal, or only represents the processing of the first aligning-heart signal and the second aligning-heart signal. That is, the stress detection method based on laser ultrasound and support vector regression provided in the embodiment of the present application can realize the present application only through the ipsilateral signal or the aligning-heart signal, and the present application will not elaborate on this one by one in the following description.

[0042] Specifically, the second preset stress may include a plurality of preset stresses σ, so as to obtain a mapping relationship between stress and delay under a plurality of different stresses, thereby improving the accuracy of stress detection.

[0043] Step S2: determining the acoustic delay characteristics of the ultrasonic signal in the composite domain based on the ultrasonic signal.

[0044] Exemplarily, step S2 is to determine the acoustic delay characteristics of the ultrasonic signal in the composite domain based on the first ultrasonic signal and the second ultrasonic signal.

[0045] Optionally, the composite domain includes the time domain, the frequency domain and the time-frequency domain; the sound delay feature includes the time domain sound delay, the frequency domain sound delay and the time-frequency domain sound delay. Step S2 includes steps S21 to S24, which are as follows: Step S21, preprocessing the ultrasonic signal to determine the time domain signal corresponding to the ultrasonic signal.

[0046] In this step, the noise reduction process of the ultrasonic signal is realized through the preprocessing step.

[0047] Exemplarily, step S21 specifically determines the acoustic delay characteristics of the ultrasonic signal in the composite domain based on the first ultrasonic signal and the second ultrasonic signal. The first time domain signal corresponds to the first frequency domain spectrum, and the second time domain signal corresponds to the second frequency domain spectrum.

[0048] Optionally, step S21 includes steps S211 to S213, which are specifically as follows: Step S211 , decomposing the ultrasonic signal based on the Symlets wavelet basis function to determine a plurality of ultrasonic sub-signals corresponding to the ultrasonic signal.

[0049] Step S212: Process each ultrasonic sub-signal based on the wavelet threshold.

[0050] Step S213: reconstructing and determining a time domain signal based on inverse wavelet transform and multiple ultrasonic sub-signals.

[0051] In this embodiment, the quality of the signal is improved and noise interference is removed. Specifically, the Symlets wavelet basis function is selected (Symlets is a wavelet basis with good symmetry, suitable for processing smooth signals), and the signal is decomposed by three layers of wavelets. On this basis, threshold processing is applied to the decomposed wavelet coefficients to suppress high-frequency noise while retaining the main features of the signal. Finally, the denoised signal is reconstructed by inverse wavelet transform, providing clearer and more accurate signal data for subsequent analysis.

[0052] Exemplarily, the first ultrasonic signal is decomposed based on the Symlets wavelet basis function to obtain three first ultrasonic sub-signals corresponding to the first ultrasonic signal, each of the first ultrasonic sub-signals is adjusted using a wavelet threshold, and then the multiple first ultrasonic sub-signals are reconstructed to form a first time domain signal. The second ultrasonic signal is decomposed based on the Symlets wavelet basis function to obtain three second ultrasonic sub-signals corresponding to the second ultrasonic signal, each of the second ultrasonic sub-signals is adjusted using a wavelet threshold, and then the multiple second ultrasonic sub-signals are reconstructed to form a second time domain signal.

[0053] Step S22, in the time domain, based on the cross-correlation analysis and the time domain signal, determine the time domain sound delay.

[0054] Optionally, step S22 includes steps S221 to S223, which are specifically as follows: Step S221: determining a first arrival time of the first ultrasonic signal and a second arrival time of the second ultrasonic signal based on the time domain signal.

[0055] In this step, since there are extreme points in ultrasonic time domain signals of different modes, the first arrival time of the first ultrasonic signal and the second arrival time of the second ultrasonic signal are determined by extracting the time corresponding to the extreme point of the time domain first wave signal.

[0056] Step S222, performing cross-correlation calculation based on the time domain signal to determine the cross-correlation peak value.

[0057] In this step, due to the inhomogeneity of stress and material microstructure, the weak dispersion characteristics of the ultrasonic signal will be caused, which will cause the waveform to be distorted and lead to errors in the acoustic delay measurement. Cross-correlation calculation can offset the acoustic delay measurement error caused by reducing signal dispersion.

[0058] Optionally, step S222 includes step S2221 and step S2222, which are specifically as follows: Perform cross-correlation calculation based on the time domain signal to determine the cross-correlation peak, including: Step S2221, based on the time domain signal, determine the cross-correlation function according to the first formula.

[0059] The first formula is as follows: First formula; in, is the cross-correlation function, To preset the observation time, is the first time domain signal, is the second time domain signal, is the time delay of the second time domain signal relative to the first time domain signal.

[0060] Step S2222, determining the time delay between signals corresponding to the maximum value of the cross-correlation function is the cross-correlation peak .

[0061] In this embodiment, The result of cross-correlation operation on the time domain signal is Related functions. Cross-correlation function The time delay corresponding to the peak Represents the maximum similarity of the signal, denoted as , the corresponding stress value is the second preset stress The time delay of the signal.

[0062] Step S223, calculating the time domain acoustic delay based on the cross-correlation peak, the first arrival time and the second arrival time.

[0063] Optionally, step S223 includes step S2231, which is specifically as follows: Step S2231, based on the cross-correlation peak value, calculate the time domain sound delay according to the second formula.

[0064] The second formula is as follows: Second formula; in, is the time domain sound delay, is the second arrival time, The first arrival time.

[0065] In this embodiment, in the time domain sound delay calculation, by subtracting The acoustic delay measurement error caused by signal dispersion can be offset.

[0066] In this embodiment, the time domain acoustic delay of the ultrasonic wave is determined based on the time domain signal. The subscript ' ' indicates time domain, superscript ' 'and' ' respectively represent the stress-free (first preset stress) state and Stress (second preset stress) state. Specifically, taking the ultrasonic signal of the additive manufacturing material with the first preset stress (i.e., the first ultrasonic signal) as a reference, the ultrasonic first wave signal when the stress is the second preset stress (i.e., the first wave signal of the second ultrasonic signal) is obtained, and the time domain acoustic delay of the first wave signal of the second ultrasonic signal relative to the first ultrasonic signal is calculated.

[0067] Step S23, in the frequency domain, determining the frequency domain sound delay based on the time domain signal.

[0068] Optionally, step S23 includes steps S231 to S236, which are specifically as follows: Step S231, based on the time domain signal, using Fourier transform technology, determine the frequency domain spectrum corresponding to the time domain signal.

[0069] In this step, when the ultrasonic signal is time-shifted in the time domain, its spectrum will produce an additional phase shift in the frequency domain, while the amplitude remains unchanged.

[0070] Exemplarily, the first time domain signal corresponds to the first frequency domain spectrum , the second time domain signal corresponds to the second frequency domain spectrum .

[0071] Step S232: performing cross-correlation analysis based on the frequency domain spectrum to determine the frequency domain spectrum cross-correlation analysis result.

[0072] Optionally, step S232 includes step S2321, which is specifically as follows: Step S2321: Based on the frequency domain spectrum, determine the frequency domain spectrum cross-correlation analysis result according to the third formula.

[0073] The third formula is as follows: The third formula; in, is the result of frequency domain spectrum cross-correlation analysis, for The conjugation of.

[0074] Step S233, determining a phase change value of the frequency domain spectrum under a second preset stress corresponding to the frequency domain spectrum cross-correlation analysis result.

[0075] Optionally, step S233 includes step S2331, which is specifically as follows: Step S2331: Based on the frequency domain spectrum cross-correlation analysis result, according to the fourth formula, determine the phase change value of the frequency domain spectrum under the second preset stress.

[0076] The fourth formula is as follows: The fourth formula; in, is the phase change value; Step S234: Calculate and determine the frequency domain acoustic delay curve based on the phase change value.

[0077] Optionally, step S234 includes step S2341, which is specifically as follows: Step S2341, based on the phase change value, determine the frequency domain sound delay curve according to the fifth formula.

[0078] The fifth formula is as follows: Fifth formula; in, is the angular frequency, , f is the frequency.

[0079] Step S235: determining a center frequency of the second time domain signal based on a second frequency domain spectrum corresponding to the second time domain signal.

[0080] Step S236: determining the frequency domain sound delay based on the frequency domain sound delay curve and the center frequency of the second time domain signal.

[0081] In this embodiment, the relationship between the frequency domain acoustic delay of the ultrasonic signal and the stress is constructed to facilitate the prediction of the stress.

[0082] Specifically, in the frequency domain sound delay calculation, the subscript 'f' in the formula represents the frequency domain, and the superscript ' 'and' ' respectively represent the stress-free (first preset stress) state and stress (second preset stress) state. In addition, , and are all functions related to the frequency f. In actual calculations, a single frequency needs to be fixed to determine the variable value corresponding to the frequency. Here, the center frequency of the signal is selected as the fixed frequency value, and the corresponding force is The frequency value corresponding to the peak of the Gaussian fitting curve is the center frequency of the signal. , then the stress is The frequency domain sound time difference is .

[0083] Step S24, in the time-frequency domain, based on the continuous wavelet transform technology and the time-domain signal, the time-frequency domain sound delay is calculated.

[0084] In view of the non-stationary broadband characteristics of laser ultrasonic signals, time-frequency analysis can comprehensively characterize the signals in both time and frequency domains, thereby more accurately revealing the dynamic characteristics of the signals. This application uses continuous wavelet transform (CWT) to analyze laser ultrasonic signals. CWT constructs wavelet basis functions by scaling and translating the mother wavelet, which can effectively capture the local characteristics of the signal in time and frequency.

[0085] Specifically, the CWT of a time domain signal is defined as:

[0086] in, represents the wavelet basis function, represents the scale parameter, represents the translation parameter, represents continuous wavelet coefficients, represents the conjugate of the wavelet basis function.

[0087] Optionally, step S24 includes steps S241 to S244, which are specifically as follows; Step S241, using continuous wavelet transform, converting the time domain signal into a time-frequency domain signal composed of wavelet coefficients.

[0088] In this step, the Complex Morlet (Cmor) wavelet is used as the wavelet basis function of the continuous wavelet transform. The choice of the wavelet basis function has a direct impact on the time-frequency analysis results. Since the Complex Morlet (Cmor) wavelet waveform is similar to the laser ultrasound waveform, and its adjustable center frequency and bandwidth parameters can significantly improve the local characteristics of the ultrasonic signal in the time-frequency domain, the Cmor wavelet is used as the wavelet basis function.

[0089] Step S242, determining the maximum value of the wavelet coefficient corresponding to the time-frequency domain signal.

[0090] In this step, the key to calculating the sound delay in the time-frequency domain is to determine the position where the signal matches the highest degree, that is, the time corresponding to the maximum value of the continuous wavelet coefficients.

[0091] Step S243, determining a reference time corresponding to the maximum value of the wavelet coefficient under the first preset stress, and a comparison time corresponding to the maximum value of the wavelet coefficient under the second preset stress.

[0092] In this step, since the time-frequency domain signal is essentially a two-dimensional matrix, the two-dimensional matrix is ​​composed of wavelet coefficients, the number of rows of the matrix is ​​determined by the frequency resolution, and the number of columns is determined by the time resolution. The wavelet time-frequency diagram corresponding to the time-frequency domain signal can be drawn using the two-dimensional matrix and the time and frequency resolution. The horizontal axis of the time-frequency diagram corresponds to the signal time, and the vertical axis corresponds to the signal frequency. By determining the maximum value of the wavelet coefficient, the signal frequency and time corresponding to the maximum value can be determined.

[0093] Step S244, calculating the time-frequency domain sound delay based on the reference time and the comparison time.

[0094] Optionally, step S244 includes step S2441, which is specifically as follows: Step S2441, based on the reference time and the comparison time, determine the time-frequency domain sound delay according to the sixth formula.

[0095] The sixth formula is as follows: Sixth formula; in, To compare the time, is the base time, is the sound delay in the time-frequency domain. Specifically, the superscript ' ' indicates the zero stress (first preset stress) state, subscript ' ' represents the time-frequency domain.

[0096] In this embodiment, the acoustic delay in the time-frequency domain is determined so as to construct a mapping relationship between stress and ultrasonic acoustic delay.

[0097] In this embodiment, the arrival time of the ultrasonic first wave signal in the time domain, frequency domain, and time-frequency domain under different stress states is analyzed and extracted to construct the changing relationship between the ultrasonic signal sound delay and stress, thereby realizing the evaluation of residual stress.

[0098] Step S3, based on the acoustic delay characteristics in the composite domain and the support vector machine regression model, a mapping relationship between the delay and the stress change is constructed.

[0099] In this step, the time domain, frequency domain and time-frequency domain acoustic delay of different modes are used as inputs of the support vector machine regression model, and the measured stress is used as the output of the support vector machine regression model. The mapping statistical law between input and output is established, and the efficiency and accuracy of online monitoring of residual stress in additive manufacturing are improved by predicting unknown stress.

[0100] It can be understood that the mapping relationship is a support vector machine regression model trained using the acoustic delay feature under the composite domain. In the case where the mapping relationship is a trained support vector machine regression model, the subsequent prediction of stress can be achieved by inputting the current ultrasonic signal into the support vector machine regression model, and the output of the support vector machine regression model obtained is the predicted stress.

[0101] To achieve stress prediction, a support vector machine regression model is used for regression analysis. Optionally, the regression function of the support vector machine regression model is In the original input space It is expressed as the following seventh formula: The seventh formula is: Seventh formula; in, represents the weight, is the bias, is the mapping function.

[0102] The support vector regression model introduces an error term , only when the absolute difference between the model output and the true value exceeds At the same time, the slack variable is introduced , , the soft margin method allows the prediction error of some training samples to exceed the set error , then the constrained optimization problem of the support vector machine regression model is shown in the eighth and ninth formulas below: Formula 8; Ninth formula; in, is the penalty factor, is the error term, , is a slack variable. Specifically, Controls the model's tolerance to training errors.

[0103] In the embodiment of the present application, the kernel function of the support vector machine regression model is a radial basis kernel function to enhance the nonlinear mapping capability.

[0104] The radial basis kernel function is shown in the following tenth formula: The tenth formula; in, is the radial basis kernel parameter, which controls the width of the kernel function and thus affects the complexity of the model. is the input of the support vector machine regression model. With the original sample input Perform operations to achieve nonlinear mapping.

[0105] Optionally, the support vector machine regression model is based on grid search to determine the optimal hyperparameter combination. In the support vector machine regression model, the penalty factor and kernel parameters These are two key hyperparameters that have a significant impact on the prediction accuracy and generalization ability of the model. This application uses grid search to optimize the parameters and obtain and The optimal parameter combination is obtained to improve the prediction accuracy of acoustic delay on stress.

[0106] Optionally, step S3 includes step S31 and step S32, which are specifically as follows: Step S31, determining a random reorganization training set based on the sound delay characteristics and random reorganization strategy in the composite domain.

[0107] Step S32: input the randomly reorganized training set into the initial support vector machine regression model, train the initial support vector machine regression model, and determine the mapping relationship.

[0108] In this embodiment, in the training process of the support vector machine regression model of the present application, in order to improve the generalization ability and prediction accuracy of the support vector regression model for laser ultrasonic signals, the acoustic delay results of different modes in the composite domain are first randomly reorganized to eliminate the deviation in the data distribution and improve the adaptability of the model. Here, in order to ensure the randomness of data partitioning, the data index generated by the random permutation function is used to generate a random order, and the random seed is fixed to ensure the repeatability of the training results. Subsequently, the reorganized data set is divided into a training set and a test set in proportion, of which 80% of the data is used for model training to ensure that the regression model fully learns the mapping relationship between stress and acoustic delay, and the remaining 20% ​​of the data is used for independent testing to evaluate the generalization ability and prediction performance of the model. Through this data partitioning strategy, the present application can effectively avoid overfitting and improve the prediction stability and accuracy of the model under different stress states.

[0109] Step S4, obtaining the current ultrasonic signal of the additive manufacturing material.

[0110] Step S5, based on the mapping relationship and in combination with the current ultrasonic signal, determining the stress value in the additive manufacturing material.

[0111] Optionally, step S5 includes step S51 and step S52, which are specifically as follows: Step S51, obtaining a current ultrasonic signal of the additive manufacturing material.

[0112] Step S52: Determine the stress value corresponding to the current ultrasonic signal in combination with the mapping relationship.

[0113] In this embodiment, by inputting the current ultrasonic signal of the additive manufacturing material into the support vector machine regression model, the output obtained is the stress value corresponding to the current ultrasonic signal, thereby achieving accurate and non-destructive detection of the stress of the additive manufacturing material.

[0114] The stress detection method based on laser ultrasound and support vector regression provided in the embodiment of the present application is adopted. By analyzing the ultrasonic signal of the additive manufacturing material under different stresses, the ultrasonic sound delay in the composite domain is determined, and then combined with the support vector regression (SVR) model, a mapping relationship between stress and ultrasonic sound delay is established, thereby realizing the prediction of unknown stress and improving the efficiency and accuracy of online stress detection in the additive manufacturing process. Long-distance, non-contact, non-destructive detection of residual stress of materials in the additive manufacturing process is realized.

[0115] Specifically, the present application uses laser ultrasonic technology to achieve long-distance, contactless stress detection of additive manufacturing materials, avoids the dependence of traditional ultrasonic detection on coupling agents and probes, and improves the adaptability to high-temperature and complex-shaped workpieces. The present application adopts time domain, frequency domain and time-frequency domain analysis methods to comprehensively extract the acoustic delay information of ultrasonic signals, thereby improving the analysis ability of ultrasonic propagation characteristics. Compared with a single analysis method, the present method can more accurately reflect the influence of stress on ultrasonic propagation characteristics and improve the stress detection accuracy. The present method achieves efficient estimation of unknown stress by constructing an acoustic delay-stress change database and using support vector regression (SVR) to establish a stress prediction model. The introduction of the SVR model effectively improves the generalization ability of the detection system, making it applicable to stress assessment under different materials and complex working conditions. By real-time detection of signals under different stress states by laser ultrasound and combining machine learning models for rapid stress prediction, the present method can achieve efficient online monitoring in the additive manufacturing process, providing real-time and effective technical support for quality control and optimization of the manufacturing process.

[0116] It is worth noting that since the acoustic elastic coefficient of the same material is the same, there is no need to construct the mapping relationship multiple times when measuring the same material. That is, after the mapping relationship is constructed and the detected additive manufacturing material is the same, there is no need to execute steps S1 to S3 again, only steps S4 and S5 are required. If the detection material is changed, the mapping relationship needs to be reconstructed because the acoustic elastic coefficient of the material changes. That is, when the detected additive manufacturing material is different, steps S1 to S5 need to be re-executed to reconstruct the mapping relationship between the stress and delay of the new material.

[0117] Figure 4 A second structural block diagram of a stress detection system based on laser ultrasound and support vector regression provided in an embodiment of the present application.

[0118] Corresponding to the aforementioned embodiment of the stress detection method based on laser ultrasound and support vector regression, the present application also provides an embodiment of a stress detection system 1 based on laser ultrasound and support vector regression. The stress detection system 1 based on laser ultrasound and support vector regression includes: a laser ultrasonic detection module 11 and a demodulation device 12. The laser ultrasonic detection module 11 is configured to collect laser ultrasonic signals for additive manufacturing materials with different stresses. The demodulation device 12 is communicatively connected to the laser ultrasonic detection module 11; the demodulation device 12 is configured to obtain ultrasonic signals of additive manufacturing materials under different stresses; based on the ultrasonic signal, determine the acoustic delay characteristics of the ultrasonic signal in the composite domain; based on the acoustic delay characteristics in the composite domain and the support vector machine regression model, construct a mapping relationship between delay and stress change; obtain the current ultrasonic signal of the additive manufacturing material; based on the mapping relationship, in combination with the current ultrasonic signal, determine the stress value in the additive manufacturing material.

[0119] The stress detection system based on laser ultrasound and support vector regression provided in the embodiment of the present application is used to analyze the ultrasonic signals of additive manufacturing materials under different stresses, determine the ultrasonic acoustic delay in the composite domain, and then combine the support vector regression (SVR) model to establish a mapping relationship between stress and ultrasonic acoustic delay, thereby realizing the prediction of unknown stress and improving the efficiency and accuracy of online stress detection in the additive manufacturing process. Long-distance, non-contact, non-destructive detection of residual stress of materials in the additive manufacturing process is realized.

[0120] Optionally, continue to refer to Figure 4 As shown, the demodulation device 12 includes a signal processing module 121 and a stress prediction module 122. The signal processing module 121 is configured to obtain ultrasonic signals of additive manufacturing materials under different stresses; based on the ultrasonic signals, determine the acoustic delay characteristics of the ultrasonic signals in the composite domain; based on the acoustic delay characteristics in the composite domain and the support vector machine regression model, construct a mapping relationship between delay and stress change. The stress prediction module 122 is configured to determine the stress value in the additive manufacturing material based on the mapping relationship and in combination with the current ultrasonic signal.

[0121] Specifically, the signal processing module 121 is used to perform multi-dimensional analysis of the acquired laser ultrasonic signals under stress-free and various stress states, including time domain analysis, frequency domain analysis, and time-frequency domain analysis, to extract the acoustic delay of the first ultrasonic wave of different modes in the composite domain. By establishing a mapping relationship between ultrasonic acoustic delay and stress, a database of acoustic delay-stress changes is constructed to provide data support for subsequent stress prediction. Based on the acoustic delay-stress change database, the stress prediction module 122 uses a support vector regression model to train the mapping relationship between acoustic delay and stress, and uses a maturely trained regression model to predict unknown stress, thereby improving the efficiency and accuracy of stress detection in the additive manufacturing process.

[0122] In a specific embodiment, in combination Figure 3 As shown, the additively manufactured material is an additively manufactured titanium alloy specimen, and this embodiment is mainly aimed at stress detection of the additively manufactured titanium alloy specimen under uniaxial tension. Figure 3 The layout of the laser excitation line source, the first point detection source (different-side centering detection method) and the second point detection source (same-side detection method) is illustrated in FIG. The stress detection process of the stress detection system based on laser ultrasound and support vector regression provided in the embodiment of the present application is as follows: In the laser ultrasonic detection module, a pulsed excitation laser is used as the ultrasonic excitation source. After the laser is beam-shaped by a collimating lens, it is finally focused into a laser line source of about 1mm (width) × 10mm (length) to excite the ultrasonic signal inside the material. In order to achieve synchronous detection of multimodal signals, a two-channel detection source is selected: the first point detection source obtains the ultrasonic signal propagating along the inside of the sample, and the second point detection source obtains the ultrasonic signal propagating along the surface of the sample. The detection laser is emitted by a continuous laser, and ultrasonic detection is performed based on the dual-wave mixing interference technology, which ensures high sensitivity and high signal-to-noise ratio of the signal.

[0123] like Figure 3 As shown, a uniaxial tensile stress of 100-350 MPa (i.e., the second preset stress) is applied to the additively manufactured titanium alloy specimen. ), and at each stress level, the ultrasound ipsilateral and concentric signals in the stress-free (first preset stress) state and different stress states are collected respectively, and the integrity and synchronization of the signal acquisition are ensured.

[0124] In order to improve the accuracy of stress prediction, it is necessary to conduct in-depth analysis of the acquired ultrasonic signals in multiple dimensions such as time domain, frequency domain, and time-frequency domain, so as to extract the corresponding acoustic delay characteristic values ​​under different stress states.

[0125] The time domain acoustic delay calculation uses the ultrasonic signal of the zero stress specimen as a reference and obtains the stress as The ultrasonic first wave signal at the time of the first wave is obtained, and its time domain acoustic delay relative to the reference signal is calculated. The time domain acoustic delay is calculated based on the arrival time difference of the two signals at the extreme point (such as the maximum peak of the first wave). In addition, due to the inhomogeneity of stress and material microstructure, the ultrasonic signal will be slightly dispersed and cause waveform distortion, which will cause errors in the acoustic delay measurement. Therefore, by performing cross-correlation calculation on the two signals, the influence of signal dispersion on the acoustic delay measurement can be reduced and the measurement accuracy can be improved.

[0126] Frequency domain acoustic delay calculation First, the ultrasonic first wave signal at the first preset stress and the second preset stress is Fourier transformed to obtain two sets of spectrum signals; then the cross-correlation Fourier transform of the two spectrum signals is calculated to obtain the phase change relationship curve of the signal under stress; then the phase change is divided by the angular frequency , and obtain the frequency sound delay variation curve; finally, the stress is The center frequency is obtained by Gaussian fitting of the signal spectrum at the time, so as to determine the frequency domain sound delay at the center frequency.

[0127] The time-frequency domain acoustic delay calculation uses continuous wavelet transform to perform time-frequency analysis on the time-domain ultrasonic signal, and selects Cmor wavelet as the wavelet basis function, with its adjustable center frequency and bandwidth parameters to improve the local characteristic performance of the signal in the time-frequency domain; in the wavelet coefficient distribution after wavelet transform, the time corresponding to the maximum coefficient value is found as the main response time of the signal; taking the signal under zero stress as the benchmark, the time-frequency domain acoustic delay under each stress state is obtained by comparing the difference in the time when the wavelet coefficient has the maximum value under different stress states.

[0128] Through the analysis of time domain, frequency domain and time-frequency domain, the acoustic delay characteristics of different modal ultrasonic first waves in these three domains are comprehensively extracted to construct an "acoustic delay-stress change database". This database provides key input data for subsequent modeling.

[0129] Based on the "acoustic delay-stress change database" constructed in the previous stage, the acoustic delay characteristics of multimodal ultrasonic signals at different stress levels were first used as model input, and the measured stress values ​​were used as model output. The support vector regression method was used for nonlinear regression modeling. Then, the penalty factor was adjusted by setting the radial basis kernel function (RBF) and combining cross-validation and grid search. and radial basis kernel parameters Optimization is performed to achieve the best generalization performance and prediction accuracy. Finally, the model is verified using data that has not participated in the training or newly measured stress states. The reliability and accuracy of the model in actual use are verified by analyzing the prediction error and stability, providing a reliable prediction model for subsequent online stress detection.

[0130] Figure 5 A schematic diagram of a process for detecting stress of additively manufactured materials provided in an embodiment of the present application.

[0131] like Figure 5 As shown in the figure, the trained and verified SVR model is integrated with the laser ultrasonic stress detection system through software and hardware to form an integrated online detection system. Its workflow includes: Step S100, the current ultrasonic signal of the additive manufacturing material.

[0132] Under an unknown stress state, laser excitation is applied to the additively manufactured titanium alloy specimen to obtain multimodal ultrasonic signals.

[0133] Step S200: extracting the acoustic delay feature of the current ultrasonic signal.

[0134] The signal processing module is used to extract the acoustic delay feature values ​​in the time domain, frequency domain and time-frequency domain.

[0135] Step S300, inputting the acoustic delay characteristics into the mapping relationship to determine the current stress value.

[0136] The above-mentioned acoustic delay feature values ​​are input into the trained SVR model (i.e., the constructed mapping relationship), the stress state of the specimen is calculated in real time, and the test results are fed back to the display or control terminal.

[0137] In summary, this application realizes non-contact, high-precision stress detection of additive manufacturing materials under complex working conditions by combining laser ultrasonic technology with support vector regression model. The comprehensive use of multimodal analysis in time domain, frequency domain and time-frequency domain effectively improves the accuracy of acoustic delay feature extraction. The constructed "acoustic delay-stress change database" and the optimized SVR model further improve the reliability and generalization ability of stress prediction. This application provides an efficient and practical residual stress detection solution in the additive manufacturing process, providing important technical support for quality control and structural health assessment.

[0138] It should be noted that those skilled in the art will easily think of other embodiments of the present application after considering the specification and practicing the application disclosed herein. The present application is intended to cover any modification, use or adaptation of the present application, which follows the general principles of the present application and includes common knowledge or customary technical means in the art that are not disclosed in the present application. The specification and examples are only regarded as exemplary, and the true scope of the present application is indicated by the claims.

[0139] It should be understood that the present application is not limited to the precise structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present application is limited only by the appended claims.

Claims

1. A stress detection method based on laser ultrasound and support vector regression, characterized in that: include: Obtain ultrasonic signals of additive manufacturing materials under different stresses; Based on the ultrasonic signal, determining the acoustic delay characteristics of the ultrasonic signal in the composite domain; Based on the acoustic delay characteristics and support vector machine regression model in the composite domain, a mapping relationship between delay and stress change is constructed; Acquiring a current ultrasonic signal of the additive manufacturing material; Based on the mapping relationship and in combination with the current ultrasonic signal, a stress value in the additive manufacturing material is determined.

2. The stress detection method based on laser ultrasound and support vector regression according to claim 1 is characterized in that: The composite domain includes a time domain, a frequency domain and a time-frequency domain; and determining the acoustic delay characteristics of the ultrasonic signal in the composite domain based on the ultrasonic signal includes: Preprocessing the ultrasonic signal to determine a time domain signal corresponding to the ultrasonic signal; In the time domain, determining a time domain acoustic delay based on a cross-correlation analysis and the time domain signal; In the frequency domain, determining a frequency domain acoustic delay based on the time domain signal; In the time-frequency domain, based on the continuous wavelet transform technology and the time-domain signal, the time-frequency domain sound delay is calculated; Among them, the sound delay characteristics include the time domain sound delay, the frequency domain sound delay and the time-frequency domain sound delay.

3. The stress detection method based on laser ultrasound and support vector regression according to claim 2 is characterized in that: The ultrasonic signal includes a first ultrasonic signal of the additive manufacturing material under a first preset stress and a second ultrasonic signal under a second preset stress; the first preset stress is 0, and the second preset stress is not 0; Determining the time domain sound delay in the time domain based on the cross-correlation analysis and the time domain signal includes: Based on the time domain signal, determining a first arrival time of the first ultrasonic signal and a second arrival time of the second ultrasonic signal; Performing cross-correlation calculation based on the time domain signal to determine a cross-correlation peak value; The time-domain acoustic delay is calculated based on the cross-correlation peak, the first arrival time, and the second arrival time.

4. The stress detection method based on laser ultrasound and support vector regression according to claim 2 is characterized in that: The ultrasonic signal includes a first ultrasonic signal of the additive manufacturing material under a first preset stress and a second ultrasonic signal under a second preset stress; the first preset stress is 0, and the second preset stress is not 0; the first time domain signal corresponds to a first frequency domain spectrum, and the second time domain signal corresponds to a second frequency domain spectrum; The step of determining, in the frequency domain, a frequency domain sound delay based on the time domain signal comprises: Based on the time domain signal, using Fourier transform technology, determine the frequency domain spectrum corresponding to the time domain signal; Based on the frequency domain spectrum, a cross-correlation analysis is performed to determine a frequency domain spectrum cross-correlation analysis result; Determine a phase change value of the frequency domain spectrum under the second preset stress corresponding to the frequency domain spectrum cross-correlation analysis result; Based on the phase change value, calculating and determining a frequency domain acoustic delay curve; Determine a center frequency of the second time domain signal based on a second frequency domain spectrum corresponding to the second time domain signal; The frequency domain sound delay is determined based on the frequency domain sound delay curve and the center frequency of the second time domain signal.

5. The stress detection method based on laser ultrasound and support vector regression according to claim 2 is characterized in that: The ultrasonic signal includes a first ultrasonic signal of the additive manufacturing material under a first preset stress and a second ultrasonic signal under a second preset stress; the first preset stress is 0, and the second preset stress is not 0; The step of calculating the time-frequency domain sound delay in the time-frequency domain based on the continuous wavelet transform technology and the time-domain signal includes: Using continuous wavelet transform, the time domain signal is converted into a time-frequency domain signal composed of wavelet coefficients; Determine the maximum value of the wavelet coefficient corresponding to the time-frequency domain signal; Determine a reference time corresponding to the maximum value of the wavelet coefficient under the first preset stress, and a comparison time corresponding to the maximum value of the wavelet coefficient under the second preset stress; Based on the reference time and the comparison time, the time-frequency domain sound delay is calculated.

6. The stress detection method based on laser ultrasound and support vector regression according to claim 2 is characterized in that: Preprocessing the ultrasonic signal to determine a time domain signal corresponding to the ultrasonic signal includes: Decomposing the ultrasonic signal based on a Symlets wavelet basis function to determine a plurality of ultrasonic sub-signals corresponding to the ultrasonic signal; Processing each of the ultrasonic sub-signals based on a wavelet threshold respectively; The time domain signal is reconstructed and determined based on using inverse wavelet transform and a plurality of the ultrasound sub-signals.

7. The stress detection method based on laser ultrasound and support vector regression according to claim 1 is characterized in that: The method of obtaining ultrasonic signals of additively manufactured materials under different stresses includes: When the additive manufacturing material has a first preset stress, collecting a first centering signal of a first point detection source and a first same-side signal of a second point detection source; When the additive manufacturing material has a second preset stress, collecting a second centering signal of the first point detection source and a second same-side signal of the second point detection source; The first preset stress magnitude is 0, and the second preset stress magnitude is non-zero.

8. The stress detection method based on laser ultrasound and support vector regression according to claim 1 is characterized in that: Based on the acoustic delay characteristics and the support vector machine regression model in the composite domain, a mapping relationship between delay and stress change is constructed, including: Determining a random reorganization training set based on the sound delay characteristics and random reorganization strategy in the composite domain; The randomly reorganized training set is input into an initial support vector machine regression model, the initial support vector machine regression model is trained, and the mapping relationship is determined.

9. The stress detection method based on laser ultrasound and support vector regression according to claim 1, characterized in that: The determining of the stress value in the additive manufacturing material based on the mapping relationship and in combination with the current ultrasonic signal includes: Acquiring a current ultrasonic signal of the additive manufacturing material; In combination with the mapping relationship, a stress value corresponding to the current ultrasonic signal is determined.

10. A stress detection system based on laser ultrasound and support vector regression, characterized in that: include: A laser ultrasonic detection module is configured to collect laser ultrasonic signals of additive manufacturing materials with different stresses; A demodulation device is communicatively connected to the laser ultrasonic detection module; the demodulation device is configured to obtain ultrasonic signals of the additive manufacturing material under different stresses; based on the ultrasonic signals, the acoustic delay characteristics of the ultrasonic signals in the composite domain are determined; Based on the acoustic delay characteristics and support vector machine regression model in the composite domain, a mapping relationship between delay and stress change is constructed; and a current ultrasonic signal of the additive manufacturing material is obtained; Based on the mapping relationship and in combination with the current ultrasonic signal, a stress value in the additive manufacturing material is determined.

Images