Concrete slab damage source positioning method and device and electronic equipment

Abstract

The invention provides a concrete slab damage source positioning method and device and electronic equipment. A sensor group is arranged on the surface of a concrete slab to capture acoustic emission signals generated by a damage source; aiming at the one-dimensional linear direction, inputting a first time difference of the acoustic emission signals received by every two sensors in the target linear direction in the sensor group into a pre-constructed linear regression model, and determining a first position coordinate corresponding to the damage source in the target linear direction; aiming at damage positioning in a two-dimensional plane, determining a second time difference of acoustic emission signals received by sensors at every two preset feature points in a sensor group in a two-dimensional space of the surface of the concrete slab, and adjusting model parameters of a pre-trained damage positioning model by taking the second time difference as training data; and inputting the second time difference into the damage positioning model after the model parameters are adjusted, and determining a second position coordinate corresponding to the damage source in the two-dimensional space on the surface of the concrete slab. And the anisotropic concrete slab can be positioned more accurately.

Description

Technical Field

[0001] The present disclosure relates to the technical field of damage location, and in particular to a method, device and electronic equipment for locating a damage source of a concrete slab. Background Art

[0002] As concrete structures are widely used in various types of construction and engineering projects, their durability and safety are particularly important. These structures are often in a changing environment and face challenges such as severe weather conditions, chemical corrosion, and periodic mechanical loads. These environmental factors can cause the gradual accumulation of structural damage, such as the development of cracks, corrosion, and other microstructural defects, which may weaken the structural integrity of concrete over time and increase safety risks. Therefore, it is essential to conduct continuous and effective health monitoring of concrete structures. This not only helps to detect potential structural problems in a timely manner and implement preventive maintenance, but also provides necessary safety warnings in emergency situations.

[0003] Acoustic emission technology is a nondestructive testing technology based on detecting tiny, instantaneous stress waves generated when materials or structures are subjected to external or internal forces. When microstructures in materials, such as cracks or holes, are active, energy is released, and the elastic waves generated during this energy release process can be captured by acoustic emission sensors. Traditional sound source localization technologies include time of arrival (TOA), time difference of arrival (TDOA), beamforming, and modal analysis. These methods usually assume that the wave velocity from the source to the sensor is constant and the propagation path is uninterrupted. However, in real concrete structures, due to the changes in structural thickness and the anisotropy of the material, the wave velocity is often not constant, which makes traditional sound source localization methods often produce large errors. Summary of the invention

[0004] The embodiments of the present disclosure at least provide a method, device and electronic equipment for locating a damage source of a concrete slab, which can more accurately locate anisotropic concrete slabs.

[0005] The present disclosure provides a method for locating a damage source of a concrete slab, comprising:

[0006] A sensor group is arranged on the surface of the concrete slab to capture the acoustic emission signal generated by the damage source;

[0007] For damage location in a one-dimensional linear direction, a first time difference between each two sensors in the sensor group in the target linear direction receiving the acoustic emission signal is input into a pre-built linear regression model to determine a first position coordinate corresponding to the damage source in the target linear direction;

[0008] For damage location in a two-dimensional plane, determine a second time difference between sensors at every two preset feature points in the sensor group in the two-dimensional space on the surface of the concrete slab receiving the acoustic emission signal, and use the second time difference as training data to adjust model parameters of a pre-trained damage location model;

[0009] The second time difference is input into the damage location model after adjusting the model parameters, and the second position coordinates corresponding to the damage source in the two-dimensional space on the surface of the concrete slab are determined.

[0010] In an optional implementation, the linear regression model is constructed based on the following steps:

[0011] Setting a sample damage source in a sample concrete slab and determining the sample damage position coordinates corresponding to the sample damage source;

[0012] Arranging a sensor group on the surface of the sample concrete slab to capture the sample acoustic emission signal generated by the sample damage source;

[0013] Determine a first sample time difference between each two sensors receiving the sample acoustic emission signal in a one-dimensional linear direction;

[0014] The linear regression model associating the first sample time difference with the sample damage position coordinates is constructed using the first sample time difference as an independent variable.

[0015] In an optional implementation, the damage localization model is trained based on the following steps:

[0016] Constructing the damage localization model with mean square error as the loss function, including a long short-term memory layer, a fully connected network and an output layer, and setting a ReLU activation function after each layer, wherein the output layer includes a horizontal axis coordinate prediction node and a vertical axis coordinate prediction node;

[0017] Determine a second sample time difference between each two sensors receiving the sample acoustic emission signal in the two-dimensional plane;

[0018] The second sample time difference and the sample damage position coordinates are input as sample data into the damage localization model, and the hyperparameters of the damage localization model are adjusted by the Bayesian optimization method so that the damage localization model has the ability to predict the position of the damage source in the two-dimensional plane according to the time difference between each sensor in receiving the acoustic emission signal.

[0019] In an optional implementation manner, using the second time difference as training data to adjust model parameters of a pre-trained damage localization model specifically includes:

[0020] In the sensor group, selecting the second time difference at which the sensor at the preset feature point receives the acoustic emission signal;

[0021] The second time difference is added to the training data corresponding to the damage localization model, and the damage localization model is continuously trained by reducing the learning rate until the damage localization model has the ability to predict the position of the damage source in the two-dimensional plane based on the time difference between each sensor receiving the acoustic emission signal on the current concrete slab.

[0022] In an optional implementation, the preset feature points are at the four corners of the concrete slab.

[0023] The embodiment of the present disclosure also provides a concrete slab damage source positioning device, comprising:

[0024] A sensor arrangement module is used to arrange a sensor group on the surface of the concrete slab to capture the acoustic emission signal generated by the damage source;

[0025] A one-dimensional linear positioning module is used for locating damage in a one-dimensional linear direction, inputting a first time difference between each two sensors in the target linear direction of the sensor group receiving the acoustic emission signal into a pre-built linear regression model, and determining a first position coordinate corresponding to the damage source in the target linear direction;

[0026] A two-dimensional plane positioning model fine-tuning module is used to determine the second time difference between the sensors at every two preset feature points in the sensor group in the two-dimensional space on the surface of the concrete slab receiving the acoustic emission signal for damage positioning in the two-dimensional plane, and use the second time difference as training data to adjust the model parameters of the pre-trained damage positioning model;

[0027] The two-dimensional plane positioning module is used to input the second time difference into the damage positioning model after adjusting the model parameters, and determine the second position coordinates corresponding to the damage source in the two-dimensional space of the concrete slab surface.

[0028] In an optional embodiment, the device is also used for:

[0029] Setting a sample damage source in a sample concrete slab and determining the sample damage position coordinates corresponding to the sample damage source;

[0030] Arranging a sensor group on the surface of the sample concrete slab to capture the sample acoustic emission signal generated by the sample damage source;

[0031] Determine a first sample time difference between each two sensors receiving the sample acoustic emission signal in a one-dimensional linear direction;

[0032] The linear regression model associating the first sample time difference with the sample damage position coordinates is constructed using the first sample time difference as an independent variable.

[0033] In an optional embodiment, the device is also used for:

[0034] Constructing the damage localization model with mean square error as the loss function, including a long short-term memory layer, a fully connected network and an output layer, and setting a ReLU activation function after each layer, wherein the output layer includes a horizontal axis coordinate prediction node and a vertical axis coordinate prediction node;

[0035] Determine a second sample time difference between each two sensors receiving the sample acoustic emission signal in the two-dimensional plane;

[0036] The second sample time difference and the sample damage position coordinates are input as sample data into the damage localization model, and the hyperparameters of the damage localization model are adjusted by the Bayesian optimization method so that the damage localization model has the ability to predict the position of the damage source in the two-dimensional plane according to the time difference between each sensor in receiving the acoustic emission signal.

[0037] The present disclosure also provides an electronic device, including: a processor, a memory and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor communicates with the memory via the bus, and when the machine-readable instructions are executed by the processor, the above-mentioned concrete slab damage source locating method or steps in any possible implementation of the above-mentioned concrete slab damage source locating method are executed.

[0038] The embodiment of the present disclosure further provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the method for locating a damage source of a concrete slab or the steps in any possible implementation of the method for locating a damage source of a concrete slab is executed.

[0039] The embodiment of the present disclosure also provides a computer program product, including a computer program / instruction, which, when executed by a processor, implements the above-mentioned concrete slab damage source locating method, or the steps in any possible implementation of the above-mentioned concrete slab damage source locating method.

[0040] The embodiment of the present disclosure provides a method, device and electronic device for locating the damage source of a concrete slab. A sensor group is arranged on the surface of the concrete slab to capture the acoustic emission signal generated by the damage source. For damage location in a one-dimensional linear direction, the first time difference between each two sensors in the sensor group in the target linear direction receiving the acoustic emission signal is input into a pre-built linear regression model to determine the first position coordinates corresponding to the damage source in the target linear direction. For damage location in a two-dimensional plane, the second time difference between each two sensors at preset feature points in the sensor group in the two-dimensional space on the surface of the concrete slab receiving the acoustic emission signal is determined, and the model parameters of the pre-trained damage location model are adjusted using the second time difference as training data. The second time difference is input into the damage location model after the model parameters are adjusted to determine the second position coordinates corresponding to the damage source in the two-dimensional space on the surface of the concrete slab. Anisotropic concrete slabs can be located more accurately.

[0041] In order to make the above-mentioned objectives, features and advantages of the present disclosure more obvious and easy to understand, preferred embodiments are specifically cited below and described in detail with reference to the attached drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the technical solutions of the embodiments of the present disclosure, the following is a brief introduction to the drawings required for use in the embodiments. The drawings herein are incorporated into the specification and constitute a part of the specification. These drawings illustrate embodiments consistent with the present disclosure and are used together with the specification to illustrate the technical solutions of the present disclosure. It should be understood that the following drawings only illustrate certain embodiments of the present disclosure and should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can also be obtained based on these drawings without creative work.

[0043] Figure 1 A flowchart of a method for locating a damage source of a concrete slab provided by an embodiment of the present disclosure is shown;

[0044] Figure 2 A flow chart showing another method for locating a damage source of a concrete slab provided by an embodiment of the present disclosure is shown;

[0045] Figure 3 A schematic diagram of a concrete slab damage source positioning device provided by an embodiment of the present disclosure is shown;

[0046] Figure 4 A schematic diagram of an electronic device provided by an embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0047] In order to make the purpose, technical scheme and advantages of the embodiments of the present disclosure clearer, the technical scheme in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the drawings in the embodiments of the present disclosure. Obviously, the described embodiments are only part of the embodiments of the present disclosure, rather than all of the embodiments. The components of the embodiments of the present disclosure generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present disclosure provided in the drawings is not intended to limit the scope of the present disclosure for protection, but merely represents the selected embodiments of the present disclosure. Based on the embodiments of the present disclosure, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present disclosure.

[0048] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0049] The term "and / or" herein only describes an association relationship, indicating that three relationships may exist. For example, A and / or B may represent the following three situations: A exists alone, A and B exist at the same time, and B exists alone. In addition, the term "at least one" herein represents any combination of at least two of any one or more of a plurality of. For example, including at least one of A, B, and C may represent including any one or more elements selected from the set consisting of A, B, and C.

[0050] Research has found that acoustic emission technology is a non-destructive testing technology based on detecting tiny, instantaneous stress waves generated when materials or structures are subjected to external or internal forces. When microstructures in materials, such as cracks or holes, are active, energy is released, and the elastic waves generated during this energy release process can be captured by acoustic emission sensors. Traditional sound source localization technologies include time of arrival (TOA), time difference of arrival (TDOA), beamforming, and modal analysis. These methods usually assume that the wave velocity from the source to the sensor is constant and the propagation path is uninterrupted. However, in real concrete structures, due to changes in structural thickness and the anisotropy of the material, the wave velocity is often not constant, which makes traditional sound source localization methods often produce large errors.

[0051] Based on the above research, the present disclosure provides a method, device and electronic device for locating the damage source of a concrete slab. A sensor group is arranged on the surface of the concrete slab to capture the acoustic emission signal generated by the damage source. For damage location in a one-dimensional linear direction, the first time difference between each two sensors in the sensor group in the target linear direction receiving the acoustic emission signal is input into a pre-built linear regression model to determine the first position coordinates corresponding to the damage source in the target linear direction. For damage location in a two-dimensional plane, the second time difference between each two sensors at preset feature points in the sensor group in the two-dimensional space on the surface of the concrete slab receiving the acoustic emission signal is determined, and the model parameters of the pre-trained damage location model are adjusted using the second time difference as training data. The second time difference is input into the damage location model after the model parameters are adjusted to determine the second position coordinates corresponding to the damage source in the two-dimensional space on the surface of the concrete slab. Anisotropic concrete slabs can be located more accurately.

[0052] To facilitate understanding of this embodiment, a concrete slab damage source location method disclosed in the embodiment of the present disclosure is first introduced in detail. The execution subject of the concrete slab damage source location method provided in the embodiment of the present disclosure is generally a computer device with certain computing capabilities, and the computer device includes, for example: a terminal device or a server or other processing device, and the terminal device can be a user equipment (User Equipment, UE), a mobile device, a user terminal, a terminal, a cellular phone, a cordless phone, a personal digital assistant (Personal Digital Assistant, PDA), a handheld device, a computing device, a vehicle-mounted device, a wearable device, etc. In some possible implementations, the concrete slab damage source location method can be implemented by a processor calling a computer-readable instruction stored in a memory.

[0053] See also Figure 1 FIG. 1 is a flowchart of a method for locating a damage source of a concrete slab provided by an embodiment of the present disclosure. The method includes steps S101 to S104, wherein:

[0054] S101. Arrange a sensor group on the surface of the concrete slab to capture the acoustic emission signal generated by the damage source.

[0055] In a specific implementation, for a concrete slab that needs to locate the damage source, a sensor group consisting of multiple sensors is arranged in an array on its surface. When the concrete slab generates tiny, instantaneous stress waves when the material or structure is subjected to external or internal forces, the microstructures in the material, such as cracks or holes, are active and release energy. The elastic waves generated during the energy release process can be captured by the acoustic emission sensor.

[0056] Here, for damage location in one-dimensional linear direction, sensors are usually arranged in a straight line or in a single direction so that the position of the damage source in this direction can be inferred by the arrival time difference. The number of sensors in the sensor group can be two, and their arrangement direction can be along the X-axis direction of the concrete slab; for damage location in a two-dimensional plane, sensors are usually arranged in multiple positions in a plane (such as different corners of a concrete slab), and the precise position of the damage source on the entire plane is calculated through time difference data in multiple directions. The number of sensors in the sensor group can be 4, set at the four corners of the concrete slab.

[0057] S102, for damage positioning in a one-dimensional linear direction, input the first time difference between each two sensors in the sensor group in the target linear direction receiving the acoustic emission signal into a pre-built linear regression model to determine the first position coordinates corresponding to the damage source in the target linear direction.

[0058] In the specific implementation, for damage location in one-dimensional linear direction, a linear regression model is used to establish a direct relationship between the time difference and the location of the damage source. One-dimensional positioning is to locate the location of the damage source only on one coordinate axis. For example, on the x-axis, the goal is to find the specific location of the damage source on this straight line. This positioning method is suitable for sensor arrays distributed along a straight line or simple linear structures.

[0059] Here, a group of sensors are arranged in one-dimensional direction (such as x-axis) of the concrete slab. When damage (such as cracks) occurs on the concrete slab, the sensor array will capture the acoustic emission signal. The time information of the sensor receiving the damage source signal is collected, and the time when the acoustic emission signal arrives at each sensor is determined by the threshold method from the signal time received by each sensor, and then the arrival time difference between each group of sensors is calculated.

[0060] Among them, a reference sensor can be selected to compare the arrival time with other sensors to obtain the arrival time difference of each group of sensors.

[0061] Furthermore, since one-dimensional positioning is linear positioning, a linear regression model is used to fit the relationship between the sensor arrival time difference and the one-dimensional coordinate value. Linear regression is a method used in statistics to predict the relationship between continuous variables, and is widely used in the fields of data analysis and machine learning. It attempts to establish a linear equation to predict the relationship between one or more independent variables (inputs) and dependent variables (outputs). It mainly includes single variable linear regression and multivariate linear regression.

[0062] Here, in the application process, the first time difference data collected in the current concrete slab is substituted into the trained linear regression model to calculate the position coordinates of the damage source in the target linear direction (eg, X-axis).

[0063] Specifically, a linear regression model is constructed based on the following steps 1-4:

[0064] Step 1: Setting a sample damage source in a sample concrete slab and determining the sample damage position coordinates corresponding to the sample damage source.

[0065] Step 2: Arrange a sensor group on the surface of the sample concrete slab to capture the sample acoustic emission signal generated by the sample damage source.

[0066] Step 3: Determine a first sample time difference between each two sensors receiving the sample acoustic emission signal in a one-dimensional linear direction.

[0067] Step 4: Taking the first sample time difference as an independent variable, construct the linear regression model that associates the first sample time difference with the sample damage position coordinates.

[0068] Here, since in the embodiment of the present application, the input is a set of arrival time differences of two sensors, and the output is the x value of the coordinate system, it belongs to single variable linear regression. The single variable linear regression model attempts to find the relationship between two variables (independent variable and dependent variable). Its mathematical expression is:

[0069] Y=β0+β1X+ε

[0070] Among them, Y is the observed value of the dependent variable, X is the observed value of the independent variable, β0 is the intercept term, β1 is the slope parameter, and ε represents the error term. When estimating the model parameters of linear regression, the most commonly used method is ordinary least squares.

[0071] In the specific implementation, the collected sample damage position coordinates and the first sample time difference are used to train the linear regression model and optimize the model parameters so that it can accurately describe the relationship between the time difference and the position. The model parameters are adjusted through the training process to minimize the prediction error, thereby improving the positioning accuracy of the model. The trained linear regression model can accurately predict the one-dimensional coordinate position.

[0072] Exemplarily, a 300mm*300mm*30mm concrete slab is used as a sample concrete slab, and a DS5 series sensor is used, with a total of four sensors (S1, S2, S3, S4). The operating frequency range of these sensors is 20kHz to 400kHz, covering the common frequency response of concrete materials, and the sampling rate is set to 3M. Coupling agent is applied to the surface of the sensor to ensure good contact with the test piece. The acoustic emission signal is amplified by a 40dB preamplifier, and the acquisition threshold is set to 100mV to effectively suppress environmental noise. To simulate the acoustic emission source, the pencil-leak break (PLB) method widely used in acoustic emission research is used, also known as the Hsu-Nielsen source. A 2B automatic pencil with a thickness of 0.5mm is used, tilted at an angle of 35° and extended with a lead core of 3mm to ensure that each break method is consistent. In the process of one-dimensional linear acoustic emission source positioning, only sensors S1 and S2 are used, with S2 as the coordinate origin, and the coordinates of S1 are (240mm, 0mm). Select (60mm, 0mm), (100mm, 0mm) and (160mm, 0mm) as training set data points, and (40mm, 0mm) and (200mm, 0mm) as test set data points. Each point in the training set breaks the lead 10 times, with a total of 30 samples. The points in the test set break the lead 10 times, with a total of 20 samples. A sensor arrival time difference corresponds to a one-dimensional coordinate x value.

[0073] S103, for damage location in a two-dimensional plane, determine a second time difference between sensors at every two preset feature points in the sensor group in the two-dimensional space on the surface of the concrete slab receiving the acoustic emission signal, and use the second time difference as training data to adjust model parameters of a pre-trained damage location model.

[0074] In the specific implementation, for the two-dimensional plane positioning problem, the long short-term memory network (LSTM) model is introduced to deeply model the relationship between the time difference and the plane position. The LSTM model is an improved recurrent neural network (RNN), which effectively solves the gradient vanishing and gradient exploding problems often encountered by traditional RNN when processing long sequence data. By introducing three core structures: input gate, forget gate and output gate, LSTM can more effectively control the storage, update and output of information in the sequence model, thereby maintaining and regulating the nonlinear characteristics of the information flow. It has been proven to show excellent performance in processing sequence data in the field of structural health monitoring.

[0075] Here, the forget gate determines which information should be remembered or forgotten, the input gate controls how information flows into the cell, and the output gate regulates the flow of information leaving the cell.

[0076] Specifically, a sensor array is arranged on the plane of the concrete slab, usually with one sensor installed at each of the four corners. When damage (such as cracks) occurs on the concrete slab, these sensors will capture the acoustic emission signal. By comparing the signal time received by each sensor, the signal arrival time difference between each sensor is calculated. Since there are four sensors, a total of six groups of arrival time differences can be formed.

[0077] In a specific implementation, the damage localization model is trained based on the following steps 1 to 3:

[0078] Step 1: construct the damage localization model with mean square error as the loss function, including a long short-term memory layer, a fully connected network and an output layer, and setting a ReLU activation function after each layer, wherein the output layer includes a horizontal axis coordinate prediction node and a vertical axis coordinate prediction node.

[0079] Step 2: Determine a second sample time difference between each two sensors receiving the sample acoustic emission signal in the two-dimensional plane.

[0080] Step 3: Input the second sample time difference and the sample damage position coordinates as sample data into the damage localization model, and adjust the hyperparameters of the damage localization model through the Bayesian optimization method so that the damage localization model has the ability to predict the position of the damage source in the two-dimensional plane according to the time difference between each sensor receiving the acoustic emission signal.

[0081] In the specific implementation, in order to establish the mapping relationship between the sensor arrival time difference and the damage location of the concrete slab, a long short-term memory network is designed as a damage location model. The input of the model is the sensor arrival time difference. A sensor is arranged at each of the four corners of the concrete slab. There are 6 groups of arrival time differences for 4 sensors, so the model input layer has 6 nodes. The core of the network is a long short-term memory (LSTM) layer containing 512 units, which aims to capture the dependencies and dynamic changes in time series data. It is followed by two layers of fully connected networks with 512 and 128 nodes respectively. The ReLU activation function is used after each layer to enhance the model's ability to handle nonlinear relationships, avoid the gradient vanishing problem, and improve the generalization ability of the model. The output layer contains two nodes, which are used to predict the x-coordinate and y-coordinate of the damage location, so the model is constructed as a regression problem.

[0082] Here, the loss function of the training model is the mean square error, and its formula is:

[0083]

[0084] Among them, loss is the loss function of the training model, MSE is the mean square error, n is the number of samples, f(x i ) and y iare the model estimated coordinate values ​​and the true coordinate values ​​respectively.

[0085] Specifically, the LSTM model is trained using known sensor arrival time difference data and corresponding damage location coordinate data. The mean square error (MSE) is used as the loss function, and the hyperparameters (such as the number of LSTM units, learning rate, batch size, and training rounds) are adjusted through the Bayesian optimization method. The model parameters are optimized through the training process so that the LSTM model can accurately map the time difference data to the plane coordinate position.

[0086] Exemplarily, a 300mm*300mm*30mm concrete slab is used as a sample concrete slab, and a DS5 series sensor is used, with a total of four sensors (S1, S2, S3, S4). The operating frequency range of these sensors is 20kHz to 400kHz, and the sampling rate is set to 3M. Coupling agent is applied to the surface of the sensor to ensure good contact with the test piece. The acoustic emission signal is amplified by a 40dB preamplifier, and the acquisition threshold is set to 100mV. To simulate the acoustic emission source, a pencil-leak break (PLB) method is used to use a 0.5mm thick 2B automatic pencil, tilted at an angle of 35° and extending 3mm of the lead core for breaking, ensuring that the method of each break is consistent. Two-dimensional planar acoustic emission source positioning is performed, the area of ​​interest is delineated on the concrete slab and a grid is constructed, the distance between the grids is 40mm, training data is collected at the 16 intersections of the grid, and test data is collected at the 9 center points of the grid. Lead breaking was performed 10 times at each intersection of the grid, and then 2 times were extracted as validation sets, and the rest were used as training sets, so the number of training samples was 128 and the number of test set samples was 32. Lead breaking was performed 5 times at the center of the grid as a test set, so the number of test samples was 45. Four sensors were deployed at the four corners of the concrete slab to receive the acoustic emission signals generated by lead breaking, and the arrival time difference from the sound source to each sensor was calculated. These time differences were used as input values ​​of the machine learning model to predict the plane coordinate values ​​x and y of the sound source. The coordinates of sensors S1, S2, S3 and S4 were set to (240mm, 0mm), (0mm, 0mm), (240mm, 240mm) and (0mm, 240mm), respectively, and a rectangular coordinate system was established with sensor S2 as the coordinate origin.

[0087] Furthermore, in order to address the problem of decreased positioning accuracy between different concrete slabs, a model fine-tuning strategy was introduced to enhance the model's adaptability to the characteristics of different concrete slabs and further improve the positioning accuracy, so as to ensure that the model can still maintain a high positioning accuracy between different slabs.

[0088] Here, model fine-tuning refers to the process of adjusting and optimizing model parameters by continuing to train on a specific task or data set based on the pre-trained model. Although all plates are made using the same manufacturing process, there are certain differences between them, which leads to a significant decrease in the positioning effect of the model when applied to other plates.

[0089] Specifically, before the model is deployed to the new slab, only the data from the four corners of each slab needs to be collected for fine-tuning, rather than repeating the complete data collection process for the initial slab. An acoustic emission signal acquisition experiment is conducted on the new concrete slab, where a sensor array is arranged and a series of acoustic emission sources at known locations are triggered (such as through artificial cracks or similar simulated signals). The arrival time of the signal received by each sensor is recorded. The signal arrival time difference between the sensors on the new slab is calculated. Since the signal propagation characteristics may change on the new slab, the time difference calculation results will be different.

[0090] As a possible implementation, see Figure 2 FIG. 1 is a flowchart of a method for locating a damage source of a concrete slab provided by an embodiment of the present disclosure. The method includes steps S1031 to S1032, wherein:

[0091] S1031. In the sensor group, select the second time difference at which the sensor at the preset feature point receives the acoustic emission signal.

[0092] S1032. Add the second time difference to the training data corresponding to the damage localization model, and continue to train the damage localization model by reducing the learning rate until the damage localization model has the ability to predict the position of the damage source in the two-dimensional plane based on the time difference between each sensor receiving the acoustic emission signal on the current concrete slab.

[0093] In the specific implementation, load the LSTM model trained on the sample plate, keep the basic structure of the model (such as the number of LSTM layers, the number of units, the nodes of the fully connected layer, etc.) unchanged, use part of the data of the new plate to fine-tune the model, and continue to train the model with a lower learning rate (such as 0.001). This can help the model gradually adjust the weight parameters on the new plate while avoiding overfitting of the old data. Most of the characteristics of the original LSTM model are retained during the training process, and only minor adjustments are made.

[0094] Here, an independent validation dataset (known damage location data not involved in training) is used to test the fine-tuned model and calculate the prediction error of the model. If the error value is large, the hyperparameters (such as learning rate, batch size, etc.) are further adjusted and training is continued.

[0095] The key to retraining the trained model is to set a lower learning rate. Such a fine-tuning strategy not only retains the original mapping relationship between the time difference and the damage location of the model, but also adapts to the specific differences of the new concrete slab, and the training resources and time will be greatly reduced. Therefore, in this case, it is crucial to maintain a low learning rate and moderate training rounds to prevent the model from over-adapting to the new data and ignoring the key information that has been learned.

[0096] Exemplarily, a 300mm*300mm*30mm concrete slab is used as a sample concrete slab, and a DS5 series sensor is used, with a total of four sensors (S1, S2, S3, S4). The operating frequency range of these sensors is 20kHz to 400kHz, and the sampling rate is set to 3M. Coupling agent is applied to the surface of the sensor to ensure good contact with the test piece. The acoustic emission signal is amplified by a 40dB preamplifier, and the acquisition threshold is set to 100mV. To simulate the acoustic emission source, the pencil-leak break (PLB) method is used. A 0.5mm thick 2B automatic pencil is used, tilted at an angle of 35° and extending 3mm of the lead core to break, ensuring that the fracture method is consistent each time. Still perform two-dimensional planar acoustic emission source positioning, and deploy the damage location model trained in the previous example to another concrete slab to test its generalization ability. In order to adapt to the characteristics of the new board, only the data of the four vertices of the area of ​​interest need to be collected as training data for fine-tuning the model, without repeating the complete data collection process of the initial board, in which each point is also cut 10 times, and a total of 40 training samples are collected. The second board is selected to cut the lead 5 times at the same test point position as the first board, and a total of 45 test samples are collected.

[0097] S104: Input the second time difference into the damage location model after adjusting the model parameters, and determine the second position coordinates corresponding to the damage source in the two-dimensional space on the surface of the concrete slab.

[0098] In a specific implementation, the second time difference between the sensors at every two preset feature points in the sensor group in the two-dimensional space of the current concrete slab surface receiving the acoustic emission signal is input into the damage localization model after the model parameters are fine-tuned, and the damage localization model after parameter fine-tuning outputs the second position coordinates (x-axis and y-axis coordinates) corresponding to the damage source in the two-dimensional space of the current concrete slab surface.

[0099] The embodiment of the present disclosure provides a method for locating a damage source of a concrete slab, wherein a sensor group is arranged on the surface of the concrete slab to capture an acoustic emission signal generated by a damage source; for damage location in a one-dimensional linear direction, a first time difference between each two sensors in the sensor group in the target linear direction receiving the acoustic emission signal is input into a pre-built linear regression model to determine a first position coordinate corresponding to the damage source in the target linear direction; for damage location in a two-dimensional plane, a second time difference between each two sensors at preset feature points in the sensor group in the two-dimensional space on the surface of the concrete slab receiving the acoustic emission signal is determined, and the model parameters of a pre-trained damage location model are adjusted using the second time difference as training data; the second time difference is input into the damage location model after the model parameters are adjusted to determine a second position coordinate corresponding to the damage source in the two-dimensional space on the surface of the concrete slab. Anisotropic concrete slabs can be located more accurately.

[0100] Those skilled in the art will appreciate that, in the above method of specific implementation, the order in which the steps are written does not imply a strict execution order and does not constitute any limitation on the implementation process. The specific execution order of the steps should be determined by their functions and possible internal logic.

[0101] Based on the same inventive concept, the embodiment of the present disclosure also provides a concrete slab damage source locating device corresponding to the concrete slab damage source locating method. Since the principle of solving the problem by the device in the embodiment of the present disclosure is similar to the above-mentioned concrete slab damage source locating method in the embodiment of the present disclosure, the implementation of the device can refer to the implementation of the method, and the repeated parts will not be repeated.

[0102] See also Figure 3 , Figure 3 Schematic diagram of a concrete slab damage source location device provided by an embodiment of the present disclosure. Figure 3 As shown in the figure, the concrete slab damage source positioning device 300 provided by the embodiment of the present disclosure includes:

[0103] The sensor arrangement module 310 is used to arrange a sensor group on the surface of the concrete slab to capture the acoustic emission signal generated by the damage source.

[0104] The one-dimensional linear positioning module 320 is used to locate damage in a one-dimensional linear direction. The first time difference between each two sensors in the target linear direction of the sensor group receiving the acoustic emission signal is input into a pre-built linear regression model to determine the first position coordinates corresponding to the damage source in the target linear direction.

[0105] The two-dimensional plane positioning model fine-tuning module 330 is used to locate damage in the two-dimensional plane, determine the second time difference between the sensors at every two preset feature points in the sensor group in the two-dimensional space on the surface of the concrete slab receiving the acoustic emission signal, and use the second time difference as training data to adjust the model parameters of the pre-trained damage localization model.

[0106] The two-dimensional plane positioning module 340 is used to input the second time difference into the damage positioning model after adjusting the model parameters, and determine the second position coordinates corresponding to the damage source in the two-dimensional space of the concrete slab surface.

[0107] For descriptions of the processing flow of each module in the device and the interaction flow between each module, reference may be made to the relevant descriptions in the above method embodiment, which will not be described in detail here.

[0108] The embodiment of the present disclosure provides a device for locating a damage source of a concrete slab. A sensor group is arranged on the surface of the concrete slab to capture the acoustic emission signal generated by the damage source. For damage location in a one-dimensional linear direction, the first time difference between each two sensors in the sensor group in the target linear direction receiving the acoustic emission signal is input into a pre-built linear regression model to determine the first position coordinates corresponding to the damage source in the target linear direction. For damage location in a two-dimensional plane, the second time difference between each two sensors at preset feature points in the sensor group in the two-dimensional space on the surface of the concrete slab receiving the acoustic emission signal is determined, and the model parameters of the pre-trained damage location model are adjusted using the second time difference as training data. The second time difference is input into the damage location model after the model parameters are adjusted to determine the second position coordinates corresponding to the damage source in the two-dimensional space on the surface of the concrete slab. Anisotropic concrete slabs can be located more accurately.

[0109] Corresponds to Figure 1 and Figure 2 The concrete slab damage source positioning method in the present disclosure also provides an electronic device 400, such as Figure 4 FIG. 4 is a schematic diagram of the structure of an electronic device 400 provided in an embodiment of the present disclosure, including:

[0110] Processor 41, memory 42, and bus 43; memory 42 is used to store execution instructions, including memory 421 and external memory 422; memory 421 here is also called internal memory, which is used to temporarily store operation data in processor 41 and data exchanged with external memory 422 such as hard disk. Processor 41 exchanges data with external memory 422 through memory 421. When the electronic device 400 is running, the processor 41 communicates with the memory 42 through bus 43, so that the processor 41 executes Figure 1 and Figure 2Steps of the concrete slab damage source location method.

[0111] The present disclosure also provides a computer-readable storage medium, which stores a computer program, and when the computer program is executed by a processor, the steps of the method for locating the damage source of a concrete slab described in the above method embodiment are executed. The storage medium can be a volatile or non-volatile computer-readable storage medium.

[0112] The present disclosure also provides a computer program product including computer instructions. When the computer instructions are executed by a processor, the steps of the method for locating the damage source of a concrete slab described in the above method embodiment can be executed. For details, please refer to the above method embodiment, which will not be repeated here.

[0113] The computer program product may be implemented in hardware, software or a combination thereof. In one optional embodiment, the computer program product is implemented as a computer storage medium. In another optional embodiment, the computer program product is implemented as a software product, such as a software development kit (SDK).

[0114] Those skilled in the art can clearly understand that, for the convenience and simplicity of description, the specific working process of the device described above can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here. In the several embodiments provided in the present disclosure, it should be understood that the disclosed device and method can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, 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 communication interfaces, and the indirect coupling or communication connection of the device or unit can be electrical, mechanical or other forms.

[0115] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0116] In addition, each functional unit in each embodiment of the present disclosure 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.

[0117] 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 non-volatile computer-readable storage medium that is executable by a processor. Based on this understanding, the technical solution of the present disclosure, 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, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present disclosure. 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.

[0118] Finally, it should be noted that the above-described embodiments are only specific implementation methods of the present disclosure, which are used to illustrate the technical solutions of the present disclosure, rather than to limit them. The protection scope of the present disclosure is not limited thereto. Although the present disclosure is described in detail with reference to the above-described embodiments, ordinary technicians in the field should understand that any technician familiar with the technical field can still modify the technical solutions recorded in the above-described embodiments within the technical scope disclosed in the present disclosure, or can easily think of changes, or make equivalent replacements for some of the technical features therein; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present disclosure, and should be included in the protection scope of the present disclosure. Therefore, the protection scope of the present disclosure should be based on the protection scope of the claims.

Claims

1. A method for locating a damage source of a concrete slab, characterized in that: include: A sensor group is arranged on the surface of the concrete slab to capture the acoustic emission signal generated by the damage source; For damage location in a one-dimensional linear direction, a first time difference between each two sensors in the sensor group in the target linear direction receiving the acoustic emission signal is input into a pre-built linear regression model to determine a first position coordinate corresponding to the damage source in the target linear direction; For damage location in a two-dimensional plane, determine a second time difference between sensors at every two preset feature points in the sensor group in the two-dimensional space on the surface of the concrete slab receiving the acoustic emission signal, and use the second time difference as training data to adjust model parameters of a pre-trained damage location model; The second time difference is input into the damage location model after adjusting the model parameters, and the second position coordinates corresponding to the damage source in the two-dimensional space on the surface of the concrete slab are determined.

2. The method according to claim 1, characterized in that The linear regression model is constructed based on the following steps: Setting a sample damage source in a sample concrete slab and determining the sample damage position coordinates corresponding to the sample damage source; Arranging a sensor group on the surface of the sample concrete slab to capture the sample acoustic emission signal generated by the sample damage source; Determine a first sample time difference between each two sensors receiving the sample acoustic emission signal in a one-dimensional linear direction; The linear regression model associating the first sample time difference with the sample damage position coordinates is constructed using the first sample time difference as an independent variable.

3. The method according to claim 2, characterized in that The damage localization model is trained based on the following steps: Constructing the damage localization model with mean square error as the loss function, including a long short-term memory layer, a fully connected network and an output layer, and setting a ReLU activation function after each layer, wherein the output layer includes a horizontal axis coordinate prediction node and a vertical axis coordinate prediction node; Determine a second sample time difference between each two sensors receiving the sample acoustic emission signal in the two-dimensional plane; The second sample time difference and the sample damage position coordinates are input as sample data into the damage localization model, and the hyperparameters of the damage localization model are adjusted by the Bayesian optimization method so that the damage localization model has the ability to predict the position of the damage source in the two-dimensional plane according to the time difference between each sensor in receiving the acoustic emission signal.

4. The method according to claim 1, characterized in that: The method of adjusting the model parameters of the pre-trained damage localization model using the second time difference as training data specifically includes: In the sensor group, selecting the second time difference at which the sensor at the preset feature point receives the acoustic emission signal; The second time difference is added to the training data corresponding to the damage localization model, and the damage localization model is continuously trained by reducing the learning rate until the damage localization model has the ability to predict the position of the damage source in the two-dimensional plane based on the time difference between each sensor receiving the acoustic emission signal on the current concrete slab.

5. The method according to claim 1, characterized in that: The preset feature points are at the four corners of the concrete slab.

6. A concrete slab damage source location device, characterized in that: include: A sensor arrangement module is used to arrange a sensor group on the surface of the concrete slab to capture the acoustic emission signal generated by the damage source; A one-dimensional linear positioning module is used to locate damage in a one-dimensional linear direction, input the first time difference between each two sensors in the target linear direction of the sensor group receiving the acoustic emission signal into a pre-built linear regression model, and determine the first position coordinates corresponding to the damage source in the target linear direction; A two-dimensional plane positioning model fine-tuning module is used to determine the second time difference between the sensors at every two preset feature points in the sensor group in the two-dimensional space on the surface of the concrete slab receiving the acoustic emission signal for damage positioning in the two-dimensional plane, and use the second time difference as training data to adjust the model parameters of the pre-trained damage positioning model; The two-dimensional plane positioning module is used to input the second time difference into the damage positioning model after adjusting the model parameters, and determine the second position coordinates corresponding to the damage source in the two-dimensional space of the concrete slab surface.

7. The device according to claim 6, characterized in that The device is also used for: Setting a sample damage source in a sample concrete slab and determining the sample damage position coordinates corresponding to the sample damage source; Arranging a sensor group on the surface of the sample concrete slab to capture the sample acoustic emission signal generated by the sample damage source; Determine a first sample time difference between each two sensors receiving the sample acoustic emission signal in a one-dimensional linear direction; The linear regression model associating the first sample time difference with the sample damage position coordinates is constructed using the first sample time difference as an independent variable.

8. The device according to claim 7, characterized in that The device is also used for: Constructing the damage localization model with mean square error as the loss function, including a long short-term memory layer, a fully connected network and an output layer, and setting a ReLU activation function after each layer, wherein the output layer includes a horizontal axis coordinate prediction node and a vertical axis coordinate prediction node; Determine a second sample time difference between each two sensors receiving the sample acoustic emission signal in the two-dimensional plane; The second sample time difference and the sample damage position coordinates are input as sample data into the damage localization model, and the hyperparameters of the damage localization model are adjusted by the Bayesian optimization method so that the damage localization model has the ability to predict the position of the damage source in the two-dimensional plane according to the time difference between each sensor in receiving the acoustic emission signal.

9. An electronic device, characterized in that: include: A processor, a memory and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor and the memory communicate via the bus, and when the machine-readable instructions are executed by the processor, the steps of the method for locating a damage source of a concrete slab as described in any one of claims 1 to 5 are performed.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method for locating a damage source of a concrete slab as claimed in any one of claims 1 to 5 are executed.

Images