A method and system for locating acoustic emission sources based on hybrid model

Abstract

The present invention provides an acoustic emission source positioning method and system based on a hybrid model, which relates to the technical field of acoustic source positioning. The method and system include: calibrating the angular range and sound speed value of each sound speed partition; calculating the theoretical reception time difference between a reference acoustic emission receiver probe and other acoustic emission receiver probes for receiving acoustic emission sources at different grid points; obtaining the actual reception time difference; calculating multiple error values ​​between the actual reception time difference and the theoretical reception time difference corresponding to the acoustic emission sources at different grid points; retaining a preset proportional error value and clustering the retained error values ​​to locate a first predicted acoustic emission source position of a target to be measured; inputting the current actual reception time difference into a random forest model to output a second predicted acoustic emission source position of the target to be measured; verifying the comprehensive confidence of the first predicted acoustic emission source position, weighting the first predicted acoustic emission source position and the second predicted acoustic emission source position, and outputting the target predicted acoustic emission source position of the target to be measured.

Description

Technical Field

[0001] The present invention relates to the technical field of sound source localization, and in particular to a method and system for localizing an acoustic emission source based on a hybrid model. Background Art

[0002] An acoustic emission source is the origin of acoustic signals generated by deformation, cracks, friction, or other physical processes within a material or structure. When a material is subjected to external forces, localized cracks expand or other deformations occur, releasing sound waves that can be detected by sensors and are called acoustic emission signals. Acoustic emission source location is determined by analyzing the propagation paths and time delays of these sound waves to pinpoint the specific location of the change.

[0003] Acoustic emission source location is crucial in many engineering fields, particularly in structural health monitoring. During use, materials or structures can develop cracks, breakage, and other problems due to external forces. Acoustic emission source location technology can monitor these potential structural defects in real time and accurately pinpoint the location of defects, helping to promptly identify problems. This effectively prevents serious damage or accidents to materials or structures during use, provides early warnings, and thus ensures the safety and reliability of engineering structures. This is particularly important in high-risk areas such as aviation, bridges, and oil pipelines. Traditional acoustic emission source location typically utilizes methods based on wave propagation theory (such as the hyperbolic location method).

[0004] However, the positioning method based on wave propagation theory often leads to positioning errors in anisotropic materials due to the non-uniformity of sound velocity, resulting in insufficient positioning accuracy. Summary of the Invention

[0005] In view of the above-mentioned deficiencies in the prior art, the purpose of the embodiments of the present invention is to provide a method for locating an acoustic emission source based on a hybrid model, which can solve the technical problems that the existing positioning method based on wave velocity propagation theory often causes positioning errors and insufficient positioning accuracy due to the non-uniformity of sound velocity in anisotropic materials.

[0006] In a first aspect of an embodiment of the present invention, a hybrid model-based acoustic emission source localization method is proposed, which is applied to a physical device for locating an acoustic emission source. The physical device for locating an acoustic emission source includes a grid plate having multiple uniform grids and multiple acoustic emission receiver probes located on the grid plate and surrounding a target to be measured. The grid plate includes multiple sound velocity zones divided at equal angles around a center point. The method includes:

[0007] S1: Calibrate the angle range and sound speed value of each sound speed partition;

[0008] S2: Based on the calibration results, calculate the theoretical receiving time difference between the reference acoustic emission receiver probe and other acoustic emission receiver probes when receiving the acoustic emission sources at different grid points;

[0009] S3: Obtain the actual receiving time difference of each acoustic emission receiver probe receiving the target acoustic emission source to be measured;

[0010] S4: Calculate multiple error values ​​between the actual receiving time difference and the theoretical receiving time difference corresponding to the acoustic emission source at different grid points by combining the grid search algorithm;

[0011] S5: retaining a preset proportion of error values, and clustering the retained error values ​​according to a clustering algorithm to locate the first predicted acoustic emission source position of the target to be measured;

[0012] S6: Use the actual receiving time difference and the corresponding grid point position as the training set to train the random forest model with multi-output regressor;

[0013] S7: Inputting the current actual receiving time difference of the target to be measured into the trained random forest model, and outputting the predicted grid point position of the target to be measured, that is, the second predicted acoustic emission source position;

[0014] S8: Verify the comprehensive confidence of the first predicted acoustic emission source position through Gaussian confidence test;

[0015] S9: Weighting the first predicted acoustic emission source position and the second predicted acoustic emission source position according to the comprehensive confidence, and outputting the target predicted acoustic emission source position of the target to be measured.

[0016] A second aspect of an embodiment of the present invention provides an acoustic emission source localization system based on a hybrid model, comprising: a processor and a memory;

[0017] The memory stores programs or instructions that can be run on the processor. When the programs or instructions are executed by the processor, the steps of the method for localizing an acoustic emission source based on a hybrid model in the first aspect are implemented.

[0018] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0019] In an embodiment of the present invention, a physical device for locating acoustic emission sources is combined with a random forest model to locate the acoustic emission sources separately. Using sound velocity zoning, the anisotropy of the material is accounted for, ensuring that the sound velocity in each region is independently calibrated. This significantly improves positioning accuracy and effectively addresses the positioning error caused by sound velocity non-uniformity in anisotropic materials using traditional wave propagation theory. A grid search and clustering algorithm are combined to optimize the positioning error, thereby accurately determining the location of the acoustic emission source. The Gaussian confidence level is then used to verify the combined confidence of the first and second predicted acoustic emission source positions. The combined confidence level is then weighted to output the target predicted acoustic emission source position of the target. Dynamically correcting the predictions from the physical and machine learning predictions in a weighted manner enhances positioning reliability and improves system accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The accompanying drawings are only for the purpose of illustrating specific embodiments and are not to be considered as limiting the present invention. Throughout the drawings, the same reference symbols represent the same components. Obviously, the drawings described below are only some embodiments of the present invention. It is clear that those skilled in the art can derive other drawings based on these drawings without inventive effort.

[0021] Figure 1 1 is a flow chart of a method for locating an acoustic emission source based on a hybrid model provided by an embodiment of the present invention;

[0022] Figure 2 This is a schematic structural diagram of a physical device for locating an acoustic emission source provided by an embodiment of the present invention;

[0023] Figure 3 is a structural diagram of a hybrid model provided by an embodiment of the present invention;

[0024] Figure 4 It is a structural diagram of an acoustic emission source localization system based on a hybrid model provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0025] In order to enable those skilled in the art to better understand the technical solutions in the embodiments of the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the embodiments described are part of the embodiments of the present invention, rather than all of the embodiments. It should be understood that these descriptions are merely exemplary and are not intended to limit the scope of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative work should fall within the scope of protection of the present invention.

[0026] The following describes in detail the acoustic emission source localization method based on the hybrid model provided by the embodiment of the present invention through specific embodiments and application scenarios in conjunction with the accompanying drawings.

[0027] Reference Manual Figure 1 , which shows a flow chart of a method for locating an acoustic emission source based on a hybrid model provided by an embodiment of the present invention.

[0028] An embodiment of the present invention provides an acoustic emission source localization method based on a hybrid model, which is applied to a physical device for locating an acoustic emission source. The physical device for locating an acoustic emission source includes a grid plate having multiple uniform grids and multiple acoustic emission receiver probes located on the grid plate and surrounding a target to be measured, wherein the grid plate includes multiple sound velocity partitions divided at equal angles with a center point.

[0029] Figure 2 The figure below illustrates an example of an optional physical device layout for AE source location. P1-P4 represent multiple AE receiver probes surrounding the target. By combining these multiple AE receiver probes, the device can capture signals from the AE source and accurately locate the source using the characteristics of different sound velocity zones. Based on actual needs, technicians can adjust the number and layout of probes to optimize location results.

[0030] Optionally, the sound speed partition can be set to 12.

[0031] In a possible implementation, the spacing between grid points of adjacent grids is 0.5 times the grid width.

[0032] The following steps may be included:

[0033] S1: Calibrate the angle range and sound speed value of each sound speed partition.

[0034] The grid plate of the acoustic emission source location device is divided into multiple sound velocity zones. Each sound velocity zone corresponds to a specific angular range. This angular range refers to the effective angle of the sound velocity zone within that zone. For example, a sound velocity zone might cover an area from 0° to 30°, meaning that all grid points within that area use the same sound velocity value. This angular range is determined based on the location of the acoustic emission source and the distribution of the receiver probes, helping the system understand the sound wave propagation characteristics of each zone. Within each sound velocity zone, the propagation speed of sound waves can vary, especially in different materials or media. The sound velocity value refers to the speed at which sound waves propagate within that specific zone and depends on the material properties and the location of the sound velocity zone. During the calibration process, a precise sound velocity value must be determined for each sound velocity zone. This value is usually derived through experimentation or theoretical calculation. The different sound velocity values ​​for different zones ensure more accurate calculation of the time difference of sound wave propagation.

[0035] It should be noted that by calibrating the angle range and sound speed value, the device can more accurately calculate the propagation path and time difference of the acoustic emission source signal, thereby improving the accuracy of source positioning.

[0036] In a possible implementation, S1 specifically includes:

[0037] S101: Initialize the angle range and sound speed value of each sound speed partition.

[0038] S102: Determine region indexes of different angles to calibrate the angle range, where the region indexes represent different angle ranges:

[0039]

[0040] Where % represents the modulo operator, g represents the region index of the angle range, θ represents the angle value, and s represents the number of sound speed partitions.

[0041] S103: Calculate the distance and relative angle between the known acoustic emission source position and each acoustic emission receiver probe.

[0042] S104: Measure the actual time difference between the known acoustic emission source position and the two different acoustic emission receiver probes.

[0043] S105: Calculate the predicted time difference between the known acoustic emission source position and the two different acoustic emission receiver probes:

[0044]

[0045] in, represents the predicted time difference between the known acoustic emission source position and the arrival time of the pth and qth acoustic emission receiver probes, d p and dq are the distances from the known acoustic emission source to the pth and qth acoustic emission receiver probes, respectively, v p and v q They represent the sound speed from the known acoustic emission source position to the pth and qth acoustic emission receiver probes respectively.

[0046] S106: Adjusting the sound speed values ​​of each sound speed partition with the goal of minimizing the time difference between the predicted time difference and the actual time difference.

[0047] S107: Perform interpolation calculation on the adjusted sound velocity value:

[0048]

[0049] Among them, v g1 and v g2 They represent the sound velocity values ​​of region indexes g1 and g2 respectively, and v represents the interpolated sound velocity value.

[0050] Specifically, first initialize the angular range and sound speed value of the sound speed partition. Next, determine the regional index (g) of different angles through S102 to calibrate the angular range, so that the sound speed partition can be calculated according to different angles. S103 calculates the distance and angle between the known sound emission source position and the receiver probe, and measures the actual time difference (S104) of the received signals of different probes. Subsequently, S105 calculates the theoretical time difference from the sound emission source to the receiver probe. The goal of S106 is to minimize the error between the predicted time difference and the actual time difference, and then adjust the sound speed value of the sound speed partition. S107 then interpolates the adjusted sound speed value, uses the L-BFGS-B algorithm to optimize, optimizes the sound speed range and outputs the final result, thereby accurately calibrating the sound speed value of each area, and optimizing the sound speed model to improve positioning accuracy.

[0051] For example, the L-BFGS-B (Limited-memory Broyden-Fletcher-Goldfarb-Shanno with Bound constraints) algorithm can be used for numerical optimization to solve large-scale, bounded, nonlinear optimization problems. The sound velocity range is set to 20 to 600 grids / milliseconds (approximately 400 to 12,000 m / s), and the sound velocity values ​​are iteratively optimized, with intermediate results output during the optimization process. The algorithm's calibrated output: After optimization is complete, the sound velocity table is updated, outputting the angular range and sound velocity values ​​for each region, for example, "Region 0 (0° to 30°): 250.3 grids / milliseconds."

[0052] S2: Based on the calibration results, calculate the theoretical receiving time difference between the reference acoustic emission receiver probe and other acoustic emission receiver probes when receiving acoustic emission sources at different grid points.

[0053] The reference acoustic emission receiver probe is a receiver probe used as a reference point in the acoustic emission source location device for comparison and calculations with other receiver probes. In a grid point acoustic emission source, the acoustic emission source is the location where the acoustic wave signal is generated, while the grid points refer to the individual subdivisions within the location area. Within the gridded location area, each grid point can be considered a potential acoustic emission source location, from which sound waves are emitted and transmitted to the receiver probe. Theoretical reception time difference: Based on the propagation speed and distance of the sound wave, the theoretical reception time difference refers to the theoretical time difference from the acoustic emission source to each receiver probe.

[0054] It should be noted that the theoretical time difference between the arrival time of sound waves emitted by the AE source at different grid points at the receiver is calculated between the reference AE receiver probe and other receiver probes. This time difference is calculated based on the theoretical model of sound wave propagation and the relative positions of different grid points, providing a basis for subsequent positioning.

[0055] In a possible implementation manner, after S2, the method further includes:

[0056] Perform a double-layer outlier cleaning on the theoretical receiving time difference. The specific cleaning process is as follows:

[0057] The theoretical receiving time differences of all grid point acoustic emission sources are preliminarily screened globally using the triple standard deviation criterion.

[0058] The theoretical receiving time difference of the acoustic emission source at all grid points is locally and finely screened using the one-standard deviation criterion.

[0059] It should be noted that, first, the theoretical reception time differences of all grid points are globally screened using the triple standard deviation criterion to remove outliers that significantly deviate from the normal range. Then, the theoretical reception time differences of each grid point are locally screened using the single standard deviation criterion to further refine the elimination of data points with large errors. These two steps effectively improve the accuracy of subsequent calculations and reduce the interference of abnormal data.

[0060] S3: Obtain the actual receiving time difference of each acoustic emission receiver probe receiving the target acoustic emission source to be measured.

[0061] It should be noted that each acoustic emission receiver probe actually receives the acoustic wave signal emitted by the target acoustic emission source, and the actual reception time difference of these signals at different receiver probes is recorded and calculated. These time differences reflect the time required for the sound wave to propagate to each receiver probe and are obtained through actual measurement. Compared with the theoretical reception time difference, the actual reception time difference better reflects the actual situation on site and is used for subsequent positioning calculations.

[0062] S4: Calculate multiple error values ​​between the actual receiving time difference and the theoretical receiving time difference corresponding to the acoustic emission source at different grid points by using a grid search algorithm.

[0063] Optionally, to implement grid search, set the grid interval to 0.5 grids, cover the detection area (e.g. 23×23 grids), and generate the coordinates of all grid points. Propagation direction calculation: For each grid point, calculate the propagation direction angle from the acoustic emission receiver probe:

[0064] θ 、 =arctan2(Δy,Δx)

[0065] Among them, θ 、 It represents the propagation direction angle from each grid point to each acoustic emission receiver probe, arctan represents the inverse tangent function, and Δy and Δx represent the coordinate difference between the grid point and the acoustic emission receiver probe.

[0066] The grid search algorithm is an optimization method that searches for the optimal solution by traversing the space of possible parameters. In acoustic emission source localization, the grid search algorithm searches for the optimal sound source location by traversing the possible locations of different grid points and calculating the error between the theoretical and actual reception time differences corresponding to these locations. The error value is the difference between the theoretical and actual reception time differences. Each grid point has an error value, which represents the difference between the actual reception time difference and the location of the acoustic emission source.

[0067] It should be noted that a grid search algorithm is used to calculate the error between the theoretical and actual reception time differences at each grid point. By setting a grid interval (e.g., 0.5 grids) to cover the entire detection area (e.g., 23 x 23 grids), the system calculates the propagation direction angle from each grid point to each acoustic emission receiver probe and uses this information to calculate the error. The goal is to compare the error between the theoretical and actual time differences to find the grid point that best approximates the actual time difference, thereby helping to accurately locate the acoustic emission source.

[0068] In a possible implementation, the error value is calculated as follows:

[0069]

[0070] in, represents the theoretical receiving time difference between the reference acoustic emission receiver probe corresponding to the i-th grid point and the k-th acoustic emission receiver probe, k = 1, 2, ..., K, K represents the total number of acoustic emission receiver probes, represents the actual receiving time difference between the reference acoustic emission receiver probe corresponding to the i-th grid point and the k-th acoustic emission receiver probe, e 0iexpress The error value between the received time and the actual receiving time.

[0071] It should be noted that the error value is calculated by comparing the theoretical reception time difference between the reference AE receiver probe and the other receiver probes with the actual reception time difference. The theoretical reception time difference is calculated using a sound speed model, while the actual reception time difference is obtained based on actual measurements. The difference between the two values ​​yields the error value. This error value reflects the deviation between the theoretical and actual reception time differences, thereby helping to optimize the location prediction of the AE source.

[0072] S5: retaining a preset proportion of error values, and clustering the retained error values ​​according to a clustering algorithm to locate a first predicted acoustic emission source position of the target to be measured.

[0073] Specifically, a certain percentage of error values ​​are first retained. These error values ​​represent the difference between the actual and theoretical reception time differences. The system then clusters these retained error values ​​using a clustering algorithm (such as DBSCAN or K-means) to help identify potential acoustic emission source locations. The point with the smallest error in the clustering results is the first predicted acoustic emission source location, which has a high degree of accuracy and is provided as the preliminary positioning result.

[0074] It should be noted that those skilled in the art can set the size of the preset ratio according to actual needs, and the present invention is not limited thereto.

[0075] In one possible implementation, the clustering algorithm is the DBSCAN algorithm. S5 specifically includes:

[0076] S501: The retained error values ​​are used as candidate points, and the candidate points are clustered using the DBSCAN algorithm.

[0077] S502: Select the minimum error value in the cluster corresponding to the maximum candidate point density.

[0078] S503: Outputting the grid point corresponding to the calculated minimum error value as the first predicted acoustic emission source position.

[0079] Specifically, in step S5, the error values ​​are first retained as candidate points and clustered using the DBSCAN algorithm. The DBSCAN algorithm clusters candidate points based on neighborhood radius and minimum sample count, ensuring that densely populated areas are accurately identified. Then, the grid point with the lowest error from the cluster with the highest density is selected as the first predicted acoustic emission source location. If no valid cluster is found, the system selects the point with the lowest error as the predicted result. This process, through clustering and error screening, improves positioning accuracy and reliability.

[0080] For example, we can filter the top 10% of candidate points for error and then perform clustering enhancement. We then cluster the candidate points using the DBSCAN algorithm (neighborhood radius eps = 1.0, minimum number of samples = 5), and select the grid point with the smallest error in the largest cluster as the predicted location. In special cases, if no valid clusters are found, we directly return the point with the smallest global error (i.e., the minimum error value among the obtained error values) to obtain the predicted location.

[0081] S6: The actual reception time difference and the corresponding grid point position are used as the training set to train the random forest model with multi-output regressor.

[0082] The random forest model is an ensemble learning model that improves prediction accuracy and robustness by constructing multiple decision trees and combining their outputs. Each decision tree uses different samples and features during training, ultimately generating a result through voting or averaging. Random forests can handle complex nonlinear problems and are highly tolerant to noise in the data. The system uses the actual reception time difference and the corresponding grid point locations as training data, using this data to train a random forest model with a multi-output regressor. The trained model can predict the specific location of the acoustic emission source based on the actual time difference. By integrating the judgments of multiple decision trees, the random forest model improves its ability to handle complex positioning problems, effectively learning from the training data and improving prediction accuracy.

[0083] In a possible implementation, S6 specifically includes:

[0084] S601: Form a feature vector from the actual receiving time differences:

[0085]

[0086] in, represents the actual receiving time difference between the reference acoustic emission receiver probe and the kth acoustic emission receiver probe, and Y represents the actual receiving time difference between the reference acoustic emission receiver probe and the kth acoustic emission receiver probe. The feature vector composed of .

[0087] S602: Normalize the feature vector:

[0088]

[0089] Among them, Y norm represents the standardized eigenvector, μ and σ represent the feature mean and feature standard deviation in the eigenvector, respectively.

[0090] S603: The normalized feature vector and the corresponding grid point position are input into a random forest for training. A multi-target regressor is used to map the feature vector into multi-target data representing the grid point position, i.e., the acoustic emission source position, until the prediction accuracy of the grid point position exceeds a preset prediction accuracy:

[0091]

[0092] in, represents the real number domain, and f represents the multi-objective regressor function.

[0093] Specifically, the system first combines all theoretical reception time differences into a feature vector, which reflects the theoretical reception time differences between the baseline acoustic emission receiver and other receivers. Then, in S602, the feature vectors are normalized by subtracting the mean and dividing by the standard deviation. This makes the feature vector values ​​more uniform and comparable, facilitating subsequent analysis. Finally, the normalized feature vectors and the corresponding grid point locations are input into a random forest model for training. The model uses a multi-target regressor to map the feature vectors to the locations of acoustic emission sources. The model is trained until the prediction accuracy reaches a preset standard. This allows the system to accurately predict the locations of acoustic emission sources based on the theoretical time differences.

[0094] It should be noted that those skilled in the art can set the preset prediction accuracy rate according to actual needs, and the present invention does not limit this.

[0095] S7: Inputting the current actual receiving time difference of the target to be measured into the trained random forest model, and outputting the predicted grid point position of the target to be measured, that is, the second predicted acoustic emission source position.

[0096] It's important to note that the actual reception time difference is fed into a trained random forest model. Based on the patterns learned during training and a comprehensive analysis of multiple decision trees, the random forest model outputs a second predicted location of the acoustic emission source. This location, predicted based on the actual reception time difference, can further verify or optimize the first predicted location, improving positioning accuracy.

[0097] S8: Verify the comprehensive confidence of the first predicted acoustic emission source position through Gaussian confidence test.

[0098] The Gaussian confidence level measures the reliability of the prediction results using a Gaussian distribution (normal distribution). It calculates the confidence level based on the residual of the reception time difference and the standard deviation of its distribution. A higher confidence level indicates a more reliable prediction result. Generally, data points with small errors and a Gaussian distribution have higher confidence levels. The reliability of the first predicted acoustic emission source location is assessed by calculating the Gaussian confidence level of the location. Specifically, the confidence level is calculated using a Gaussian distribution based on the residual between the theoretical reception time difference and the actual reception time difference. If the Gaussian confidence level of the first predicted location is high, it indicates that the prediction result is relatively reliable and can be further used to weightedly fuse other prediction results.

[0099] In a possible implementation, S8 specifically includes:

[0100] S801: Calculate the theoretical receiving time differences at the first predicted acoustic emission source position:

[0101] S802: Calculate the reception time residual between each theoretical reception time difference and the actual reception time difference at the first predicted sound emission source position.

[0102] S803: Calculate the Gaussian confidence of the residual at each receiving moment:

[0103]

[0104] Among them, Φ() represents the Gaussian density function, It represents the receiving time residual between the reference acoustic emission receiver probe and the kth acoustic emission receiver probe, that is, the difference between the theoretical receiving time difference and the current actual receiving time difference, C 1k It represents the Gaussian confidence of the residual error in the reception time between the reference acoustic emission receiver probe and the kth acoustic emission receiver probe, and τ represents the standard deviation parameter.

[0105] Optionally, the standard deviation parameter can be set to 0.1.

[0106] S804: Take the average of each Gaussian confidence level to obtain the comprehensive confidence level:

[0107]

[0108] Among them, C 综合 Indicates the overall confidence.

[0109] Specifically, the system first calculates the theoretical reception time differences at the first predicted AE source location. It then calculates the residuals between these theoretical time differences and the actual reception time differences. Next, the system uses a Gaussian density function to calculate the Gaussian confidence level for each residual, representing the reliability of each residual at each reception time. Finally, all Gaussian confidence levels are averaged to produce a composite confidence level. This composite confidence level reflects the overall reliability of all predicted locations and helps determine the accuracy of the prediction results. A higher composite confidence level indicates a more reliable predicted AE source location.

[0110] S9: Weighting the first predicted acoustic emission source position and the second predicted acoustic emission source position according to the comprehensive confidence, and outputting the target predicted acoustic emission source position of the target to be measured.

[0111] It's important to note that the first and second predicted AE source locations are weighted based on their combined confidence levels. Locations with higher overall confidence levels are given greater weight, ensuring a more accurate final predicted target location. This weighting method allows the system to combine the two different predictions to generate a more comprehensive and reliable final AE source location, thereby improving positioning accuracy.

[0112] In a possible implementation, S9 specifically includes:

[0113] S901: When the comprehensive confidence is greater than the preset comprehensive confidence, weight the first predicted sound emission source position and the second predicted sound emission source position according to the comprehensive confidence, and proceed to step 903; otherwise, proceed to step S902.

[0114] S902: Weighting the first predicted sound emission source position and the second predicted sound emission source position according to a preset fixed weight value.

[0115] S903: Calculate and output the target predicted acoustic emission source position. The target predicted acoustic emission source position is calculated as follows:

[0116]

[0117] Among them, ε represents the preset fixed weight value, C represents the intermediate variable representing the weight value, C0 represents the preset comprehensive confidence, X phy and X ml Represent the first predicted acoustic emission source position and the second predicted acoustic emission source position, respectively. Indicates the target predicted acoustic emission source position.

[0118] It should be noted that those skilled in the art can set the preset fixed weight value and the preset comprehensive confidence level according to actual needs, and the present invention does not limit this.

[0119] Optionally, the preset fixed weight value may be set to 0.01.

[0120] Specifically, if the combined confidence level is greater than a preset value, the system weights the first and second predicted AE source locations based on the combined confidence level to determine the target predicted AE source location. Otherwise, a fixed weighting value is used for the weighted calculation. In this calculation, C is an intermediate variable in the weighting process, and the combined confidence level determines the weights used in the weighting process, ultimately resulting in the target predicted AE source location. This weighting approach allows the system to combine the two prediction results, ensuring higher accuracy.

[0121] Reference Manual Figure 3 , which shows a structural diagram of a hybrid model provided by an embodiment of the present invention.

[0122] Figure 3 This solution illustrates an example of a modular hybrid model structure. The data input layer is used to obtain the actual reception time difference between each acoustic emission receiver probe and the acoustic emission source of the target under test. The data preprocessing module is used to perform a two-layer outlier cleanup on the theoretical reception time difference. The physical model branch, namely the acoustic emission source positioning physical device, is used to locate the first predicted acoustic emission source position of the target under test. The machine learning branch, namely the random forest model, is used to output the second predicted acoustic emission source position of the target under test. The model fusion layer verifies the comprehensive confidence of the first predicted acoustic emission source position using the Gaussian confidence criterion, and weights the first and second predicted acoustic emission source positions based on the comprehensive confidence to output the target predicted acoustic emission source position of the target under test.

[0123] In actual application, the device first calculates the theoretical reception time difference between the reference receiver and other receivers by calibrating the sound speed partitions. Then, the actual reception time difference is obtained and the error value is calculated using a grid search algorithm. This is then combined with a clustering algorithm to optimize the positioning results. Next, a random forest model is trained using the theoretical time difference and grid point locations as training data to predict the second location of the acoustic emission source. Finally, the Gaussian confidence levels of the two predicted locations are comprehensively considered and the results are combined using a weighted method to output the final, accurate acoustic emission source location. This entire process combines traditional physical models with modern machine learning techniques to improve positioning accuracy and reliability.

[0124] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0125] In an embodiment of the present invention, a physical device for locating acoustic emission sources is combined with a random forest model to locate the acoustic emission sources separately. Using sound velocity zoning, the anisotropy of the material is accounted for, ensuring that the sound velocity in each region is independently calibrated. This significantly improves positioning accuracy and effectively addresses the positioning error caused by sound velocity non-uniformity in anisotropic materials using traditional wave propagation theory. A grid search and clustering algorithm are combined to optimize the positioning error, thereby accurately determining the location of the acoustic emission source. The Gaussian confidence level is then used to verify the combined confidence of the first and second predicted acoustic emission source positions. The combined confidence level is then weighted to output the target predicted acoustic emission source position of the target. Dynamically correcting the predictions from the physical and machine learning predictions in a weighted manner enhances positioning reliability and improves system accuracy.

[0126] Reference Manual Figure 4 , which shows a structural schematic diagram of an acoustic emission source localization system based on a hybrid model provided by an embodiment of the present invention.

[0127] The embodiment of the present invention provides an acoustic emission source localization system 20 based on a hybrid model, comprising: a processor 201 and a memory 202;

[0128] The memory 202 stores a program or instruction that can be run on the processor 201. When the program or instruction is executed by the processor 201, the steps of the above-mentioned hybrid model-based acoustic emission source localization method are implemented, and the same technical effect can be achieved. To avoid repetition, the present invention will not be repeated.

[0129] It should be understood that the processor 201 in the embodiment of the present invention may be a central processing unit (CPU), and the processor may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc.

[0130] It should also be understood that the memory 202 in the embodiment of the present invention can be a volatile memory or a non-volatile memory, or can include both volatile and non-volatile memories. Among them, the non-volatile memory can be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory can be a random access memory (RAM), which is used as an external cache. By way of example and not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic random access memory (DRAM), synchronous DRAM (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link DRAM (SLDRAM), and direct rambus RAM (DR RAM).

[0131] The above embodiments can be implemented in whole or in part through software, hardware (such as circuits), firmware, or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer program are loaded or executed on a computer, the processes or functions described in accordance with the embodiments of the present invention are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via a wired method (such as infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that contains a collection of one or more available media. The available medium can be a magnetic medium (such as a floppy disk, hard disk, or magnetic tape), an optical medium (such as a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state drive.

[0132] It should be understood that in various embodiments of the present invention, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0133] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present invention.

[0134] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described equipment, devices and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0135] In the several embodiments provided by the present invention, it should be understood that the disclosed devices, apparatuses and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interface, indirect coupling or communication connection of the device or unit, which can be electrical, mechanical or other forms.

[0136] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0137] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0138] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0139] An embodiment of the present invention provides a readable storage medium including: a program or instruction stored on the readable storage medium, and when the program or instruction is executed by a processor, the steps of the above-mentioned method for localizing an acoustic emission source based on a hybrid model are implemented, and the same technical effect can be achieved. To avoid repetition, the present invention will not be described in detail.

[0140] Finally, it should be noted that the above embodiments are merely illustrative of the technical solutions of the embodiments of the present invention, and are not intended to limit them. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they may still modify the technical solutions described in the aforementioned embodiments, or replace some of the technical features therein with equivalents; and such modifications or replacements do not deviate from the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention. Any changes or replacements that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be covered by the scope of protection of the present invention.

Claims

1. A method for locating acoustic emission sources based on a hybrid model, characterized in that: The invention is applied to a physical device for locating an acoustic emission source, the physical device comprising a grid plate having a plurality of uniform grids and a plurality of acoustic emission receiver probes located on the grid plate and surrounding a target to be measured, wherein the grid plate comprises a plurality of sound velocity zones divided at equal angles with respect to a center point; the method comprising: S1: calibrating the angle range and sound speed value of each sound speed partition; S2: Based on the calibration results, calculate the theoretical receiving time difference between the reference acoustic emission receiver probe and other acoustic emission receiver probes when receiving the acoustic emission sources at different grid points; S3: obtaining the actual receiving time difference of each acoustic emission receiver probe receiving the target acoustic emission source to be measured; S4: Calculating multiple error values ​​between the actual receiving time difference and the theoretical receiving time difference corresponding to the acoustic emission sources at different grid points by using a grid search algorithm; S5: retaining a preset proportion of error values, and clustering the retained error values ​​according to a clustering algorithm to locate the first predicted acoustic emission source position of the target to be measured; S6: Using the actual receiving time difference and the corresponding grid point position as a training set to train a random forest model with a multi-output regressor; S7: Inputting the current actual receiving time difference of the target to be measured into the trained random forest model, and outputting the predicted grid point position of the target to be measured, that is, the second predicted acoustic emission source position; S8: verifying the comprehensive confidence of the first predicted acoustic emission source position by using Gaussian confidence test; S9: Weighting the first predicted acoustic emission source position and the second predicted acoustic emission source position according to the comprehensive confidence, and outputting the target predicted acoustic emission source position of the target to be measured.

2. The method for locating an acoustic emission source based on a hybrid model according to claim 1, wherein: The spacing between grid points of adjacent grids is 0.5 times the grid width.

3. The method for locating an acoustic emission source based on a hybrid model according to claim 1, wherein: Said S1 specifically includes: S101: Initializing the angle range and sound speed value of each sound speed partition; S102: Determine region indexes of different angles to calibrate the angle range, wherein the region indexes represent different angle ranges: Among them, % represents the modulo operator, g represents the region index of the angle range, θ represents the angle value, and s represents the number of sound speed partitions; S103: Calculating the distance and relative angle between the known acoustic emission source position and each of the acoustic emission receiver probes; S104: measuring the actual time difference between the known acoustic emission source position and the two different acoustic emission receiver probes; S105: Calculate the predicted time difference between the known acoustic emission source position and the two different acoustic emission receiver probes: in, represents the predicted time difference between the known acoustic emission source position and the arrival time of the pth and qth acoustic emission receiver probes, d p and d q are the distances from the known acoustic emission source to the pth and qth acoustic emission receiver probes, respectively, v p and v q Respectively represent the sound speed from the known acoustic emission source position to the pth and qth acoustic emission receiver probes; S106: adjusting the sound speed value of each sound speed partition with the goal of minimizing the time difference between the predicted time difference and the actual time difference; S107: Perform interpolation calculation on the adjusted sound velocity value: Among them, v g1 and v g2 They represent the sound velocity values ​​of region indexes g1 and g2 respectively, and v represents the interpolated sound velocity value.

4. The method for locating an acoustic emission source based on a hybrid model according to claim 1, wherein: After S2, the method further includes: The theoretical receiving time difference is cleaned of double-layer outliers. The cleaning process is as follows: The theoretical receiving time difference of all grid point acoustic emission sources is preliminarily screened globally using the triple standard deviation criterion. The theoretical receiving time differences of the acoustic emission sources at all grid points are locally and finely screened using a one-time standard deviation criterion.

5. The method for locating an acoustic emission source based on a hybrid model according to claim 1, wherein: The error value is calculated as follows: in, represents the theoretical receiving time difference between the reference acoustic emission receiver probe corresponding to the i-th grid point and the k-th acoustic emission receiver probe, k = 1, 2, ..., K, K represents the total number of acoustic emission receiver probes, represents the actual receiving time difference between the reference acoustic emission receiver probe corresponding to the i-th grid point and the k-th acoustic emission receiver probe, e 0i express The error value between the received time and the actual receiving time.

6. The method for locating an acoustic emission source based on a hybrid model according to claim 1, characterized in that: The clustering algorithm is the DBSCAN algorithm; S5 specifically includes: S501: taking the retained error values ​​as candidate points and clustering the candidate points using the DBSCAN algorithm; S502: Select the minimum error value in the cluster corresponding to the maximum candidate point density; S503: Outputting the calculated grid point corresponding to the minimum error value as the first predicted acoustic emission source position.

7. The method for locating an acoustic emission source based on a hybrid model according to claim 1, wherein: The S6 specifically includes: S601: Combining the actual receiving time differences into a feature vector: in, represents the actual receiving time difference between the reference acoustic emission receiver probe and the kth acoustic emission receiver probe, and Y represents the actual receiving time difference between the reference acoustic emission receiver probe and the kth acoustic emission receiver probe. The characteristic vector composed of S602: Normalize the feature vector: Among them, Y norm represents the standardized eigenvector, μ and σ represent the feature mean and feature standard deviation in the eigenvector respectively; S603: The normalized feature vector and the corresponding grid point position are input into the random forest for training. The feature vector is mapped into multi-target data representing the grid point position, i.e., the acoustic emission source position, using a multi-target regressor until the prediction accuracy of the grid point position exceeds a preset prediction accuracy: in, represents the real number domain, and f represents the multi-objective regressor function.

8. The method for locating an acoustic emission source based on a hybrid model according to claim 5, characterized in that: The S8 specifically includes: S801: Calculate the theoretical receiving time differences at the first predicted acoustic emission source position: S802: Calculating the receiving time residual between each theoretical receiving time difference and the actual receiving time difference at the first predicted sound emission source position; S803: Calculate the Gaussian confidence of the residuals at each receiving moment: Among them, Φ() represents the Gaussian density function, It represents the receiving time residual between the reference acoustic emission receiver probe and the kth acoustic emission receiver probe, that is, the difference between the theoretical receiving time difference and the current actual receiving time difference, C 1k represents the Gaussian confidence of the residual error in the reception time between the reference acoustic emission receiver probe and the kth acoustic emission receiver probe, and τ represents the standard deviation parameter; S804: Taking the average of the Gaussian confidences to obtain the comprehensive confidence: Among them, C 综合 Indicates the overall confidence.

9. The method for locating an acoustic emission source based on a hybrid model according to claim 1, wherein: The S9 specifically includes: S901: If the comprehensive confidence is greater than a preset comprehensive confidence, weight the first predicted sound emission source position and the second predicted sound emission source position according to the comprehensive confidence, and proceed to step 903; otherwise, proceed to step S902; S902: Weighting the first predicted sound emission source position and the second predicted sound emission source position according to a preset fixed weight value; S903: Calculate and output the target predicted acoustic emission source position. The target predicted acoustic emission source position is calculated as follows: Among them, ε represents the preset fixed weight value, C represents the intermediate variable representing the weight value, C0 represents the preset comprehensive confidence, X phy and X ml Represent the first predicted acoustic emission source position and the second predicted acoustic emission source position, respectively. Indicates the target predicted acoustic emission source position.

10. An acoustic emission source localization system based on a hybrid model, characterized in that: include: processor and memory; The memory stores programs or instructions that can be run on the processor, and when the programs or instructions are executed by the processor, the steps of the acoustic emission source localization method based on the hybrid model according to any one of claims 1 to 9 are implemented.

Images