Crack identification method based on acoustic emission, medium and terminal
By training a GMM model using the EM algorithm and combining posterior probability and confidence index, crack classification is automatically performed, solving the problems of cumbersome operation and low accuracy in existing technologies, and realizing convenient and efficient crack identification.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SUN YAT SEN UNIV
- Filing Date
- 2025-09-29
- Publication Date
- 2026-05-12
AI Technical Summary
Existing crack identification methods based on acoustic emission are cumbersome to operate, have low accuracy, lack credibility evaluation and mechanism-driven quantitative means, and are difficult to deploy and interpret quickly in engineering practice.
The GMM model is trained using the EM algorithm. By extracting the RA and AF parameters of the AE signal, the posterior probability is calculated. The crack identification confidence index Ci and the crack mechanism energy dominance index are introduced to automatically classify cracks.
It simplifies the classification process, improves identification accuracy and interpretability, is applicable to engineering practice, and enables convenient crack identification.
Smart Images
Figure CN122020349A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of crack monitoring technology, and particularly relates to a crack identification method, medium and terminal based on acoustic emission. Background Technology
[0002] Acoustic emission (AE) technology, as a non-destructive testing method, is widely used in the health monitoring of materials and structures. AE events often reflect crack mechanisms through characteristic parameters (such as RA and AF values), thereby distinguishing between shear cracks and tensile cracks. Gaussian mixture models (GMMs) are soft clustering algorithms capable of modeling complex distributions and providing posterior probabilities. They are commonly used for AE parameter clustering analysis. However, GMMs only provide the cluster structure and cannot directly classify cracks; additional discrimination boundaries need to be constructed or new methods introduced for discrimination.
[0003] Existing crack classification methods based on acoustic emission (AE) characteristic parameters rely on manual boundary judgment or additional discrimination methods (such as Support Vector Machines, SVMs; Bayesian Information Criterion, BIC) for the classification process, resulting in complex classification steps, low accuracy, and poor generalization ability. Furthermore, existing methods lack credibility evaluation of classification results and mechanism-driven quantification methods, failing to provide clear and interpretable auxiliary judgment indicators for practical engineering applications. Patent application CN116821726A proposes a UHPC tensile-shear crack discrimination method based on acoustic emission characteristic parameter clustering analysis. First, it uses acoustic emission to synchronously monitor the UHPC damage and fracture process to obtain acoustic emission characteristic parameters. Then, it calculates RA and AF values based on these parameters to construct a label-free dataset. Secondly, a Gaussian mixture model (GMM) clustering algorithm is used to calculate the probability of each sample point being generated from different clusters. This process is iterated multiple times to update the cluster probabilities until Gaussian convergence. Each sample is then assigned to the cluster with the highest probability, resulting in two clusters. The sample point data is then labeled. The labeled sample dataset is presented with RA values on the x-axis and AF values on the y-axis. A support vector machine (SVM) algorithm is used to find the optimal linear boundary between the two clusters, which is then used to distinguish between tensile and shear cracks. This patent application uses an SVM discriminant method. SVM relies on pseudo-labels output by the GMM, increasing the model's structural complexity. Furthermore, this boundary method lacks clear physical meaning, stability, and automated classification capabilities, hindering rapid deployment and interpretation in practical engineering, and exhibiting the same drawbacks as existing technologies.
[0004] Therefore, how to provide a crack identification method that is easy to operate and has high accuracy is a problem that urgently needs to be solved by those in this technical field. Summary of the Invention
[0005] In view of the shortcomings of the prior art, the purpose of this invention is to provide a crack identification method based on acoustic emission, so as to solve the problems of cumbersome operation and low accuracy in the crack identification process of the prior art; in addition, this invention also provides a crack identification medium and terminal based on acoustic emission.
[0006] To solve the above-mentioned technical problems, the present invention adopts the following technical solution:
[0007] In a first aspect, the present invention provides a crack identification method based on acoustic emission, comprising the following steps:
[0008] S10. Extract the RA and AF parameters of the AE signal and construct a two-dimensional feature set;
[0009] S20. Train the GMM model using the EM algorithm and output the mean, covariance matrix and weight of each Gaussian distribution component.
[0010] S30. Determine the posterior probability of each AE event in all Gaussian distributions;
[0011] S40. Compare the posterior probabilities of each component and take the category corresponding to the maximum value as the classification result of the crack type of the event.
[0012] S50, Introducing the Crack Identification Confidence Index C i Select threshold C th To determine the result, when C i ≥C th If Ci < C, then the classification result of the event is reliable. th In such cases, the classification results should be used with caution;
[0013] S60. The crack mechanism energy dominance index is introduced to determine the degree of energy dominance of the shear mechanism during failure, thus assisting in engineering analysis.
[0014] Furthermore, the GMM model expression in step S20 is as follows:
[0015]
[0016] Where M is the number of single Gaussian models, and ωk is the weight of each single Gaussian model.
[0017] N(x|u k ,Σ k Let Σ be the probability density function of the Kth SGM. k Let u be the covariance matrix. k As expected.
[0018] Furthermore, the EM algorithm is used for parameter estimation to construct the objective function J(θ) to its maximum value, and the mean, covariance matrix, and weight coefficients are initialized:
[0019] θ=[ω i ,u i ,Σ i ], i = 1, 2, ..., M;
[0020]
[0021] ω i =1 / M.
[0022] Furthermore, the EM algorithm is divided into an expectation step and a maximization step. In the expectation step, the posterior probability of each Gaussian component is calculated as follows:
[0023]
[0024] In the maximization step, the model parameters are updated based on the calculated posterior probabilities as follows:
[0025]
[0026]
[0027] Repeat the expectation step and the maximization step until the convergence condition is met, and the difference between the two likelihood estimates satisfies the preset value, i.e., |J(θ). t+1 -J(θ) t |<ε.
[0028] Furthermore, in step S50, C i =|P 1i -P 2i |, threshold C th It is 0.0002.
[0029] Furthermore, in step S40, the posterior probability Pk(x) output by the GMM model is used, where k = 1 and 2 are two Gaussian components. For each AE event, x... i =[RA i ,AF i Calculate the probability that it belongs to one of the two distributions:
[0030] P1=w1·N(x i |μ1,Σ1);
[0031] P2=w2·N(x i |μ2,Σ2);
[0032] If P1 > P2, it is classified as component 1; otherwise, it is classified as component 2; if they are equal, it is determined to be a mixed crack.
[0033] Furthermore, the expression in step S60 is as follows:
[0034]
[0035] Among them, E i Let be the energy of the i-th AE event, and let Pshear,i and Ptensile,i be the posterior probabilities of belonging to shear crack and tensile crack, respectively. 拉伸 i, whose classification confidence level is C i =|P cut, iP 拉伸 ,i|,C th As the confidence threshold, only C i ≥C th Only events that are considered "unacceptable" are included in the statistics.
[0036] In a second aspect, the present invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the method described above.
[0037] Thirdly, the present invention also provides an electronic terminal, comprising: a processor and a memory; the memory is used to store a computer program, and the processor is used to execute the computer program stored in the memory to cause the terminal to perform the method described above.
[0038] Compared with existing technologies, the crack identification method, medium, and terminal based on acoustic emission provided by this invention have at least the following advantages:
[0039] Existing crack classification methods based on acoustic emission (AE) feature parameters rely on manual boundary judgment or additional discrimination methods (such as Support Vector Machines, SVMs, and Bayesian Information Criterion, BIC) for classification, resulting in complex classification steps, low accuracy, and poor generalization ability. This invention offers a simple and convenient process, requiring no manual annotation or intervention. It introduces an automatic classification strategy based on the posterior probability of Geometric Matrix (GMM), classifying cracks according to the clustering results of the GMM method, thus improving recognition accuracy. Furthermore, by constructing a credibility evaluation index and a crack dominance index, it enhances the interpretability and engineering applicability of crack identification. Attached Figure Description
[0040] To more clearly illustrate the solution of the present invention, a brief introduction will be given to the drawings used in the description of the embodiments below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.
[0041] Figure 1A flowchart illustrating a crack identification method based on acoustic emission, provided as an embodiment of the present invention;
[0042] Figure 2 This is a schematic diagram of the existing GMM output results;
[0043] Figure 3 This is a schematic diagram of the classification results from existing classification methods.
[0044] Figure 4 A schematic diagram of the GMM output result of a crack identification method based on acoustic emission provided in an embodiment of the present invention;
[0045] Figure 5 A schematic diagram of the classification results of a crack identification method based on acoustic emission provided in an embodiment of the present invention;
[0046] Figure 6 This is a schematic diagram of the classification results of a crack identification method based on acoustic emission provided in an embodiment of the present invention, taking a compression test of a certain material at different loading rates as an example. Detailed Implementation
[0047] To facilitate understanding of the present invention, a more complete description will be given below with reference to the accompanying drawings. Preferred embodiments of the invention are shown in the drawings. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided to provide a thorough and complete understanding of the disclosure of the invention.
[0048] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention.
[0049] This invention provides a crack identification method based on acoustic emission, applied in the monitoring of materials and structures. Specifically, in the crack identification process, the crack identification method based on acoustic emission includes:
[0050] S10. Extract the RA and AF parameters of the AE signal and construct a two-dimensional feature set; S20. Train the GMM model using the EM algorithm and output the mean, covariance matrix and weight of each Gaussian distribution component.
[0051] S30. Determine the posterior probability of each AE event in all Gaussian distributions; S40. Compare the posterior probabilities of each component and take the category corresponding to the maximum value as the classification result of the crack type of the event; S5. Introduce the crack identification confidence index C. i Select threshold C th To determine the result, when Ci ≥C th If Ci < C, then the classification result of the event is reliable. th In such cases, the classification results should be used with caution; S60, introduce the energy dominance index of crack mechanism to judge the degree of energy dominance of shear mechanism during failure, and assist in engineering analysis.
[0052] The present invention has a simple process, is easy to operate, and effectively improves the accuracy of crack identification.
[0053] To enable those skilled in the art to better understand the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings.
[0054] This invention provides a crack identification method based on acoustic emission, applied in the monitoring of materials and structures, specifically in the crack identification process, combined with... Figures 1 to 6 , Figure 2 and Figure 3 The existing GMM output diagrams and classification results from existing classification methods cannot directly distinguish between tensile cracks and shear cracks, making the judgment difficult and unclear. Figure 4 and Figure 5 Mesh 1 represents the Gaussian distribution of shear cracks, and Mesh 2 represents the Gaussian distribution of tensile cracks. The two distributions show a significant difference in intensity, allowing for the determination of which type of crack has a higher probability. The result is simple, direct, clear, and straightforward. Specifically, in this embodiment, the crack identification method based on acoustic emission includes:
[0055] S10. Extract the RA and AF parameters of the AE signal and construct a two-dimensional feature set.
[0056] S20. Train the GMM model using the EM algorithm and output the mean, covariance matrix and weight of each Gaussian distribution component.
[0057] Specifically, in this embodiment, the GMM model expression is as follows:
[0058]
[0059] Where M is the number of single Gaussian models, and ωk is the weight of each single Gaussian model.
[0060] N(x|u k ,Σ k Let Σ be the probability density function of the Kth SGM. k Let u be the covariance matrix. k As expected.
[0061] Furthermore, in this embodiment, to obtain the optimal values of the GMM parameters, the EM algorithm is used for parameter estimation, constructing an objective function J(θ) to reach its maximum value, and initializing the mean, covariance matrix, and weight coefficients:
[0062] θ=[ω i ,u i ,Σ i ], i = 1, 2, ..., M;
[0063]
[0064] ω i =1 / M.
[0065] Furthermore, in this embodiment, the Expectation-Maximization (EM) algorithm is used to estimate the parameters of the Gaussian Mixture Model (GMM). The EM method consists of an expectation step and a maximization step. In the expectation step, the posterior probability of each Gaussian component is calculated as follows:
[0066]
[0067] In the maximization step, the model parameters are updated based on the calculated posterior probabilities as follows:
[0068]
[0069] Repeat the expectation step and the maximization step until the convergence condition is met, and the difference between the two likelihood estimates satisfies the preset value, i.e., |J(θ). t+1 -J(θ) t |<ε.
[0070] S30. Determine the posterior probability of each AE event in all Gaussian distributions.
[0071] S40. Compare the posterior probabilities of each component and take the category corresponding to the maximum value as the classification result of the crack type of the event.
[0072] Specifically, in this embodiment, the posterior probability Pk(x) output by the GMM model is used, where k = 1 and 2 are two Gaussian components. For each AE event, x i =[RA i ,AF i Calculate the probability that it belongs to one of the two distributions:
[0073] P1=w1·N(x i |μ1,Σ1);
[0074] P2=w2·N(x i |μ2,Σ2);
[0075] If P1 > P2, it is classified as component 1 (shear crack); otherwise, it is classified as component 2 (tensile crack); if they are equal, it is classified as a mixed crack.
[0076] S50, Introducing the Crack Identification Confidence Index C i Select threshold C th To determine the result, when C i ≥C th If Ci < C, then the classification result of the event is reliable. th In such cases, the classification results should be used with caution.
[0077] Specifically, in this embodiment, C i =|P 1i -P 2i | where P is the posterior probability, P 1i P represents the probability of a shear crack. 2i The threshold C represents the probability of tensile cracking. th It can be selected as 0.0002.
[0078] S60. The crack mechanism energy dominance index is introduced to determine the degree of energy dominance of the shear mechanism during failure, thus assisting in engineering analysis.
[0079] Specifically, in this embodiment, the expression is as follows:
[0080]
[0081] Among them, E i Let be the energy of the i-th AE event, and let Pshear,i and Ptensile,i be the posterior probabilities of belonging to shear crack and tensile crack, respectively. 拉伸 i, whose classification confidence level is C i =|P cut, iP 拉伸 ,i|,C th As the confidence threshold, only C i ≥C th Only events that are considered "unacceptable" are included in the statistics.
[0082] Example 1
[0083] Taking a compression test of a material at different loading rates as an example, the material specimen is a cylindrical specimen with a diameter of 50 mm and a height of 100 mm. The loading method is displacement control, and the loading rates are 0.0001, 0.001, 0.01, and 0.1 mm / min. A DS-8B acoustic emission instrument is used to collect the elastic wave signal during the specimen's failure process. The acoustic emission acquisition frequency is set to 3 MHz, the threshold is set to 10 mV, and the amplifier gain is set to 40 dB. The method provided in this embodiment of the invention is used to determine the material failure mode. The results for the 0.1 mm / min loading condition are as follows: Figure 4 and Figure 5As shown, acoustic emission events can be clearly and easily classified. The coverage areas of shear cracks and tensile cracks are relatively clear. Furthermore, the energy dominance index R-value of the crack mechanism under four working conditions is calculated, and the results are as follows: Figure 6 As shown, with the increase of loading rate, R 剪切 R continues to increase 拉伸 The continuously decreasing results indicate a shift in crack classification for this type of material under different strain rates, which is clearer than the classification by existing technologies.
[0084] This invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any of the methods in this embodiment.
[0085] This invention also provides an electronic terminal, including: a processor and a memory; the memory is used to store a computer program, and the processor is used to execute the computer program stored in the memory, so that the terminal performs any of the methods in this embodiment.
[0086] As will be understood by those skilled in the art, the computer-readable storage medium described in this embodiment allows for the implementation of all or part of the steps in the above method embodiments by computer program-related hardware. The aforementioned computer program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.
[0087] The electronic terminal provided in this embodiment includes a processor, a memory, a transceiver, and a communication interface. The memory and the communication interface are connected to the processor and the transceiver and complete communication between them. The memory is used to store computer programs, the communication interface is used to perform communication, and the processor and the transceiver are used to run the computer programs, so that the electronic terminal performs the steps of the above method.
[0088] Compared with existing technologies, the above-described crack identification method, medium, and terminal based on acoustic emission (AE) feature parameters suffer from problems such as complex classification steps, low accuracy, and poor generalization ability. Existing crack classification methods based on AE feature parameters rely on manual boundary judgment or additional discrimination methods (such as Support Vector Machines (SVM) or Bayesian Information Criterion (BIC)). This invention features a simple and convenient process, requiring no manual annotation or intervention. It introduces an automatic classification strategy based on the posterior probability of Generative Models (GMM), classifying cracks according to the clustering results of the GMM method, thus improving identification accuracy. Furthermore, by constructing a credibility evaluation index and a crack dominance index, it enhances the interpretability and engineering applicability of crack identification.
[0089] Obviously, the embodiments described above are merely preferred embodiments of the present invention, and not all embodiments. The accompanying drawings illustrate preferred embodiments of the present invention, but do not limit the scope of the patent. The present invention can be implemented in many different forms; rather, these embodiments are provided to provide a more thorough and complete understanding of the disclosure of the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing specific embodiments, or make equivalent substitutions for some of the technical features. Any equivalent structures made using the content of this specification and drawings, directly or indirectly applied to other related technical fields, are similarly within the scope of patent protection of this invention.
Claims
1. A crack identification method based on acoustic emission, characterized in that, Includes the following steps: S10. Extract the RA and AF parameters of the AE signal and construct a two-dimensional feature set; S20. Train the GMM model using the EM algorithm and output the mean, covariance matrix and weight of each Gaussian distribution component. S30. Determine the posterior probability of each AE event in all Gaussian distributions; S40. Compare the posterior probabilities of each component and take the category corresponding to the maximum value as the classification result of the crack type of the event. S50, Introducing the Crack Identification Confidence Index C i Select threshold C th To determine the result, when C i ≥C th If Ci < C, then the classification result of the event is reliable. th In such cases, the classification results should be used with caution; S60. The crack mechanism energy dominance index is introduced to determine the degree of energy dominance of the shear mechanism during failure, thus assisting in engineering analysis.
2. The crack identification method based on acoustic emission according to claim 1, characterized in that, The GMM model expression in step S20 is as follows: ； ； Where M is the number of single Gaussian models, and ωk is the weight of each single Gaussian model. , Let K be the probability density function of the Kth SGM. Let covariance matrix be the variance matrix. As expected.
3. The crack identification method based on acoustic emission according to claim 2, characterized in that, The EM algorithm is used for parameter estimation, and the objective function is constructed. To maximize this value, initialize the mean, covariance matrix, and weight coefficients: ； ； 。 4. The crack identification method based on acoustic emission according to claim 3, characterized in that, The EM algorithm consists of an expectation step and a maximization step. In the expectation step, the posterior probability of each Gaussian component is calculated as follows: ； In the maximization step, the model parameters are updated based on the calculated posterior probabilities as follows: ； ； ； Repeat the expectation step and the maximization step until the convergence condition is met, and the difference between the two likelihood estimates satisfies the preset value, i.e. .
5. The crack identification method based on acoustic emission according to claim 4, characterized in that, In step S50 Threshold C th It is 0.0002.
6. The crack identification method based on acoustic emission according to claim 5, characterized in that, In step S40, the posterior probability Pk(x) output by the GMM model is used, where k=1,2 are two Gaussian components. For each AE event... Calculate the probability that it belongs to one of the two distributions: ； ； like If the crack is equal to a certain value, it is classified as component 1; otherwise, it is classified as component 2; if they are equal, it is determined to be a mixed crack.
7. The crack identification method based on acoustic emission according to claim 6, characterized in that, The expression in step S60 is as follows: ； Among them, E i Let be the energy of the i-th AE event, whose posterior probabilities of belonging to shear crack and tensile crack are Pshear,i and Ptension,i, respectively. Its classification confidence is Ci = |Pshear,iPtension,i|, and Cth is the confidence threshold. Only events with Ci ≥ Cth are included in the statistics.
8. A computer-readable storage medium, characterized in that, The storage medium stores a computer program that, when executed by a processor, implements the method as described in any one of claims 1 to 7.
9. An electronic terminal, characterized in that, include: Processor and memory; The memory is used to store a computer program, and the processor is used to execute the computer program stored in the memory to cause the terminal to perform the method as described in any one of claims 1 to 7.