A damage identification method based on lamb wave temperature compensation and related device

Abstract

The present application belongs to the technical field of train safety monitoring, and particularly relates to a damage identification method based on Lamb wave temperature compensation and related devices. The damage identification method comprises the following processes: obtaining a Lamb wave signal and an actual ambient temperature of a component to be measured; inputting the Lamb wave signal and the actual ambient temperature into a pre-constructed Lamb wave temperature compensation network model to output a temperature-compensated Lamb wave signal; and identifying damage of the component to be measured based on the temperature-compensated Lamb wave signal to obtain a damage identification result. After temperature compensation, temperature-related interference is removed, so that damage identification based on the compensated Lamb wave signal is more accurate.

Description

Technical Field

[0001] This invention belongs to the field of train safety monitoring technology, and in particular relates to a damage identification method and related device based on Lamb wave temperature compensation. Background Technology

[0002] Structural health monitoring is crucial for maintaining the integrity of engineering structures. In recent years, Lamb waves have attracted widespread attention due to their applications in structural health monitoring. These waves possess advantages such as long propagation distances and sensitivity to structural damage, demonstrating great potential in monitoring the integrity of plate-like structures. However, Lamb waves themselves exhibit frequency dispersion and multi-mode characteristics; they are complex combinations of boundary echoes, damage-induced reflections, and noise, making the extraction of damage information from these signals a challenge. In Lamb wave-based structural health monitoring techniques, researchers primarily focus on the differences between signals in healthy and damaged states, analyzing these differences to extract damage scattering characteristics, thereby enabling the detection and assessment of structural damage.

[0003] In engineering practice, the performance of Lamb waves is susceptible to fluctuations in ambient temperature. Existing research has shown that even a tiny temperature change of only 0.5°C can lead to significant variations in Lamb wave amplitude. Temperature fluctuations not only alter the piezoelectric coefficient of piezoelectric sensors but also affect the mechanical properties of materials, such as the elastic modulus of aluminum alloys. This results in a decrease in Lamb wave amplitude as temperature rises, along with a corresponding increase in wave propagation time. Given the extensive coverage and vast geographical span of high-speed railway networks, train structures inevitably encounter variable climatic conditions. These variable environmental conditions not only pose challenges to the structural safety of high-speed trains but can also severely interfere with Lamb wave monitoring signals, making accurate identification of damage signals difficult and increasing the risk of misjudgment.

[0004] In summary, traditional damage monitoring methods can only be used to monitor damage in a single-temperature environment, do not take into account the influence of temperature on the Lamb wave signal, and are not suitable for monitoring damage in variable-temperature environments. Summary of the Invention

[0005] The purpose of this invention is to provide a damage identification method and related device based on Lamb wave temperature compensation, which solves the problem that existing Lamb wave damage detection methods are not suitable for monitoring damage in variable temperature environments.

[0006] This invention is achieved through the following technical solution: A damage identification method based on Lamb wave temperature compensation includes the following process: Acquire the Lamb wave signal and actual ambient temperature of the component under test; The Lamb wave signal and the actual ambient temperature are input into a pre-built Lamb wave temperature compensation network model, and the temperature-compensated Lamb wave signal is output. Damage identification is performed on the component under test based on the temperature-compensated Lamb wave signal, and the damage identification results are obtained.

[0007] Furthermore, the construction process of the pre-built Lamb wave temperature compensation model is as follows: Cracks are fabricated on the target structure of the component to be tested according to the preset crack size; A sensor is installed at the target structure to acquire Lamb wave signals of different damage sizes on the component under test at multiple different ambient temperatures as raw data; the multiple different ambient temperatures include room temperature and very low temperature. The original data is truncated to obtain the basic symmetrical modes of the Lamb wave signal, and then normalized to serve as sample data. The sample data are labeled with temperature based on the ambient temperature to obtain a sample set; the sample set is then divided into a training set and a test set. A Lamb wave temperature compensation network is constructed based on a conditional generative adversarial network; the Lamb wave temperature compensation network includes a generator and a discriminator. The training set is input into the Lamb wave temperature compensation network for adversarial training until the generator and discriminator reach Nash equilibrium. The test set is input into the trained Lamb wave temperature compensation network, which outputs the temperature-compensated Lamb wave signal until the test is passed, thus obtaining the constructed Lamb wave temperature compensation network model.

[0008] Furthermore, the test set is input into the trained Lamb wave temperature compensation network, which outputs the temperature-compensated Lamb wave signal until the test is passed, thus obtaining the constructed Lamb wave temperature compensation network model, specifically: The test set is input into the trained Lamb wave temperature compensation network, which outputs the temperature-compensated Lamb wave signal. The temperature-compensated Lamb wave signal is used to identify the target structural cracks in the component under test and output the predicted damage size. The error between the predicted damage size and the preset crack size is calculated until the error between the predicted damage size and the preset crack size meets the requirements. Then the Lamb wave temperature compensation network model is completed.

[0009] Furthermore, the process of inputting the training set into the Lamb wave temperature compensation network for adversarial training is as follows: Train the discriminator: The original input signal and temperature label are input into the generator, which generates a reconstructed signal. The reconstructed signal and the target real signal are input into the discriminator, and the output result is fed back to the discriminator through the loss function to complete the update of the discriminator parameters; during this process, the generator parameters are frozen. Train the generator: With the parameters of the discriminator fixed, the original input signal and temperature label are input into the generator, and the generator generates the reconstructed signal. The reconstructed signal and the target real signal are input into the discriminator, and similarity is evaluated through a loss function. If the similarity evaluation meets the preset requirements, training stops; if the similarity evaluation does not meet the preset requirements, the similarity loss is backpropagated to the generator, and the generator parameters are iteratively updated until the similarity evaluation meets the preset requirements, at which point the generator training is complete. The generator and discriminator are trained alternately until Nash equilibrium is reached; The original input signal is the Lamb wave signal acquired at a very low ambient temperature, and the target real signal is the Lamb wave signal acquired at room temperature.

[0010] Furthermore, damage identification is performed on the component under test based on the temperature-compensated Lamb wave signal, specifically as follows: The temperature-compensated Lamb wave signal is input into the damage detection network, and the predicted damage result is output.

[0011] Furthermore, the damage detection network employs a convolutional neural network.

[0012] This invention discloses a damage identification system based on Lamb wave temperature compensation, comprising the following processes: The data acquisition module is used to acquire the Lamb wave signal and actual ambient temperature of the component under test; The temperature compensation module is used to input the Lamb wave signal and the actual ambient temperature into the pre-built Lamb wave temperature compensation network model and output the temperature-compensated Lamb wave signal. The damage identification module is used to identify damage to the component under test based on the temperature-compensated Lamb wave signal and obtain the damage identification result.

[0013] The present invention also discloses a computer device, including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor executes the computer program to implement the steps of the damage identification method based on Lamb wave temperature compensation.

[0014] The present invention also discloses a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the damage identification method based on Lamb wave temperature compensation.

[0015] The present invention also discloses a computer program product, including a computer program / instruction that, when executed by a processor, implements the steps of the damage identification method based on Lamb wave temperature compensation.

[0016] Compared with the prior art, the present invention has the following beneficial technical effects: This invention discloses a damage identification method based on Lamb wave temperature compensation. A temperature compensation model is constructed, and this method considers the influence of temperature on Lamb wave signals. Temperature changes alter the physical properties of materials, such as elastic modulus and density, thus affecting the propagation characteristics of Lamb waves and introducing temperature-related interference into the original Lamb wave signal. Temperature compensation removes this temperature-related interference, making damage identification based on the compensated Lamb wave signal more accurate. In test components with significant temperature variations in different seasons or working environments, it can more reliably detect damage size, avoiding misjudgments or omissions caused by temperature changes. Furthermore, by acquiring Lamb wave signals at different temperatures on the target structure of the test component, this method is applicable to various test components with similar structures, possessing a certain degree of versatility and can be extended to damage detection in different fields.

[0017] Furthermore, when constructing the Lamb wave temperature compensation model, the original data was truncated to obtain the basic symmetric modes as sample data, and temperature labels were assigned based on the ambient temperature. This processing method effectively extracted key information related to damage and temperature, reduced redundant components in the data, improved data quality, and facilitated subsequent model training and learning.

[0018] A Lamb wave temperature compensation network is constructed based on Conditional Generative Adversarial Networks (CGAN). The generator and discriminator are trained adversarially to learn the mapping relationship of Lamb wave signals at different temperatures. Alternating training of the generator and discriminator until Nash equilibrium is reached ensures that the reconstructed signal generated by the generator approximates the target real signal (Lamb wave signal at room temperature), thus achieving effective compensation for Lamb wave signals under extremely high temperatures. This adversarial training mechanism has strong learning ability and adaptability, and can capture the complex influence of temperature changes on Lamb wave signals. Attached Figure Description

[0019] Figure 1 This is a diagram illustrating the adversarial training process of a temperature compensation model according to the present invention. Figure 2 This refers to the aluminum plate 6061 experimental piece and the overall experimental setup in Embodiment 3 of the present invention; Figure 3This is a schematic diagram of the Lamb wave signal in Embodiment 3 of the present invention; Figure 4 A structural diagram of the generator provided by this invention; Figure 5a The Lamb wave signals are obtained under non-destructive conditions at temperatures of -30°C and room temperature. Figure 5b The Lamb wave signals are given at a damage size of 5 mm and temperatures of -30°C and room temperature. Figure 5c The Lamb wave signals are given at a damage size of 4.5 mm and temperatures of -10℃ and room temperature. Figure 5d The Lamb wave signals are given at a damage size of 5 mm, a temperature of 70 °C, and a normal temperature. Figure 5e The Lamb wave signals are given at a damage size of 3 mm, a temperature of 50 °C, and a normal temperature. Figure 5f The corresponding Lamb wave signals are given at a damage size of 7 mm, a temperature of -10℃, and a normal temperature. Figure 6a The performance of uncompensated Lamb wave signals in damage prediction and damage diagnosis; Figure 6b This invention demonstrates the performance of a damage identification method based on Lamb wave temperature compensation in damage diagnosis. Figure 7 This is a flowchart of a damage identification method based on Lamb wave temperature compensation according to the present invention; Figure 8 This is a block diagram of a damage identification system based on Lamb wave temperature compensation according to the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of the present invention clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention; that is, the described embodiments are only a part of the embodiments of the present invention, and not all of them.

[0021] The components described and illustrated in the accompanying drawings and embodiments of this invention can be arranged and designed in various different configurations. Therefore, the detailed description of the embodiments of the invention provided in the following drawings is not intended to limit the scope of the claimed invention, but merely to illustrate one selected embodiment of the invention. All other embodiments obtained by those skilled in the art based on the accompanying drawings and embodiments of this invention without inventive effort are within the scope of protection of this invention.

[0022] It should be noted that the terms “comprising,” “including,” or any other variations are intended to cover non-exclusive inclusion, such that a process, element, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to the process, element, method, article, or apparatus.

[0023] This invention provides a damage identification method based on Lamb wave temperature compensation, which can detect damage in complex service environments.

[0024] The features and performance of the present invention will be further described in detail below with reference to embodiments.

[0025] Example 1 like Figure 7 As shown, this invention discloses a damage identification method based on Lamb wave temperature compensation, comprising the following steps: Acquire the Lamb wave signal and actual ambient temperature of the component under test; The Lamb wave signal and the actual ambient temperature are input into a pre-built Lamb wave temperature compensation network model, and the temperature-compensated Lamb wave signal is output. Damage identification is performed on the component under test based on the temperature-compensated Lamb wave signal, and the damage identification results are obtained.

[0026] Example 2 Based on Example 1, please refer to Figure 1 This embodiment provides a method for constructing a Lamb wave temperature compensation network model, including the following process: Cracks are artificially created on the target structure of the component to be tested according to preset dimensions; N sensors are set at the target structure of the component under test; the sensors acquire Lamb wave signals of different damage sizes on the component under test at multiple different ambient temperatures as raw data; the multiple different ambient temperatures include room temperature and extreme ambient temperature; Then, a fixed-size sliding window is used to truncate the original data to obtain the basic symmetric mode of the Lamb wave signal, namely the S0 mode, as sample data; The sample data are labeled with temperature based on the ambient temperature to obtain a sample set; the sample set is then divided into a training set and a test set. Construct a Lamb wave temperature compensation network based on conditional generative adversarial networks; The training set is input into the Lamb wave temperature compensation network for adversarial training until the generator and discriminator reach Nash equilibrium. The test set is input into the trained Lamb wave temperature compensation network, which outputs the temperature-compensated Lamb wave signal until the test is passed, thus obtaining the constructed Lamb wave temperature compensation network model.

[0027] Specifically, the original data is truncated using a sliding window of fixed size to obtain the basic symmetrical modes of the Lamb wave signal, and then normalized according to the following formula.

[0028]

[0029]

[0030] in, The signal is after normalization. For signal The average value; Let n be the number of signal sampling points, where n is the nth sampling point.

[0031] Specifically, the Lamb wave temperature compensation network includes a generator and a discriminator. For example... Figure 4 As shown, the designed generator consists of an encoder, a transformer, and a decoder. The encoder and decoder can be viewed as inverse operations, while the transformer mainly acts as a transition to mitigate gradient vanishing.

[0032] The encoder and decoder architecture consists of multiple trainable convolutional layers used to extract features from the input Lamb wave signal. These convolutional layers include convolutionals, normalization instances, and activation functions. At each convolutional layer, all kernels are convolved with the input and a bias is added. The features are then transformed using the activation function, as shown in the following expression.

[0033]

[0034] In the formula, for Layer The output of each convolutional feature map. "Indicates convolution operation, Indicates in The i-th neuron in layer and Layer Weights between feature maps; This is the activation function.

[0035] The converter consists of multiple residual modules, which are designed to connect the encoder and decoder. The main purpose is to alleviate the gradient vanishing and gradient exploding problems that occur during the data mapping process.

[0036] The discriminator and generator use a discriminative loss function formed by their own outputs. Alternately update the parameters of the discriminator and the generator.

[0037] The adversarial training process consists of two steps: First, the discriminator D is trained: the original input signal is used as the training input. The temperature label y is input into the generator, which generates a reconstruction signal. Reconstructed signal and target true signal The input is fed into the discriminator D, which distinguishes between the reconstructed input signal and the target real signal, constructs the cross-entropy loss, and optimizes the parameters of the discriminator; during this process, the parameters of the generator G are frozen. The original input signal is the Lamb wave signal acquired at a very low ambient temperature, and the real signal is the Lamb wave signal acquired at room temperature. Secondly, the generator should be trained: With the parameters of the discriminator D fixed, the original input signal... The temperature label y is input into the generator, which generates a reconstruction signal. Reconstructed signal and target true signal Similarity is evaluated using a loss function. If the similarity evaluation meets the preset requirements, training stops. If the similarity evaluation does not meet the preset requirements, the similarity loss is backpropagated to the generator, causing the generator's parameters to be updated iteratively until the similarity evaluation meets the preset requirements, at which point the generator training is complete.

[0038] The generator G and discriminator D are trained alternately until Nash equilibrium is reached.

[0039] in,

[0040]

[0041] To reconstruct the signal, This represents maximizing the loss function with a fixed discriminator D and minimizing the loss function with a fixed generator G; y is the temperature label.

[0042] The reconstructed signal is obtained through generator G. Subsequently, a Lamb wave signal similarity measurement method is proposed, using this similarity as a loss to train and optimize the entire Lamb wave temperature compensation network. The similarity between the reconstructed signal and the target real signal is calculated, and the calculation process is as follows:

[0043] Where n is the number of samples. and The target real signal and the reconstructed signal are represented respectively. By optimizing this loss function, the gap between the generated network and the target real signal can be gradually reduced, achieving temperature compensation. The entire optimization process updates the parameters in the Lamb wave temperature compensation network using the Adam optimization algorithm. The weights of this network are trained from the data itself and have adaptive characteristics. Therefore, it can adapt well to Lamb wave temperature compensation under varying temperatures.

[0044] The test set is input into the trained Lamb wave temperature compensation network, which outputs the temperature-compensated Lamb wave signal of the test set. Then, the temperature-compensated Lamb wave signal is input into the damage detection network, which outputs the predicted damage size. The error between the predicted damage size and the preset crack size is calculated until the error between the predicted damage size and the preset crack size meets the requirements. Then, the Lamb wave temperature compensation network model is completed.

[0045] Damage detection networks can utilize existing convolutional neural networks, which can effectively capture local features and spatial information in Lamb wave signals, resulting in high accuracy in damage detection.

[0046] Example 3 This embodiment describes a specific experimental verification method, as follows: Twenty-one 2mm thick 6061 aluminum plates, each 400mm x 200mm in size, were used as experimental specimens. Their mechanical parameters are shown in Table 1. Different damage sizes were pre-set on each aluminum plate, ranging from 0 to 10mm in 0.5mm intervals. The specific structural design and sensor placement are as described above. Figure 2 Similarly, a programmable chamber is used to change the ambient temperature to simulate the service environment of the actual structure.

[0047] Table 1: Mechanical parameters of 6061 aluminum plate

[0048] Overall experimental setup as follows Figure 2 As shown, a programmable temperature chamber is used to change the ambient temperature to simulate the service environment of the actual structure. The simulated temperature range is -40℃ to 80℃. After a preset temperature is maintained stably for 10 minutes, one sensor is used as an exciter, and the other sensor receives the Lamb wave signal. This process is repeated for each aluminum plate, and the Lamb signal waveforms received by the sensors at different ambient temperatures are recorded, as shown below. Figure 3 As shown. In this way, a large amount of data was obtained as raw data.

[0049] Then, a fixed-size sliding window is used to truncate the original data to obtain the basic symmetric mode of the Lamb wave signal, namely the S0 mode, and normalize it as sample data. The sample data are labeled with temperature based on the ambient temperature to obtain a sample set; the sample set is then divided into a training set and a test set. The training set is input into the Lamb wave temperature compensation network for adversarial training until the generator and discriminator reach Nash equilibrium. The test set is input into the optimized Lamb wave temperature compensation network, and the output is the temperature-compensated Lamb wave signal. This process continues until the test is passed, and the constructed Lamb wave temperature compensation network model is obtained.

[0050] like Figure 5a As shown, the original input signal at -30℃ without damage is input into the constructed Lamb wave temperature compensation network, and the output is the temperature-compensated Lamb wave signal, which is the reconstructed signal. Simultaneously, the actual target signal at 20℃ without damage is input as a control.

[0051] like Figure 5b As shown, the original input signal corresponding to a temperature of -30℃ and a damage size of 5mm is input into the constructed Lamb wave temperature compensation network, and the output is the temperature-compensated Lamb wave signal, which is the reconstructed signal. Simultaneously, the actual target signal corresponding to a temperature of 20℃ and a damage size of 5mm is also input as a control.

[0052] like Figure 5c As shown, the original input signal corresponding to a temperature of -10℃ and a damage size of 4.5mm is input into the constructed Lamb wave temperature compensation network, and the output is the temperature-compensated Lamb wave signal, which is the reconstructed signal. Simultaneously, the actual target signal corresponding to a temperature of 20℃ and a damage size of 4.5mm is also input as a control.

[0053] like Figure 5d As shown, the original input signal corresponding to a temperature of 70℃ and a damage size of 5mm is input into the constructed Lamb wave temperature compensation network, and the output is the temperature-compensated Lamb wave signal, which is the reconstructed signal. Simultaneously, the actual target signal corresponding to a temperature of 20℃ and a damage size of 5mm is input as a control.

[0054] like Figure 5e As shown, the original input signal corresponding to a temperature of 50℃ and a damage size of 3mm is input into the constructed Lamb wave temperature compensation network, and the output is the temperature-compensated Lamb wave signal, which is the reconstructed signal. Simultaneously, the actual target signal corresponding to a temperature of 20℃ and a damage size of 3mm is also input as a control.

[0055] like Figure 5fAs shown, the original input signal corresponding to a temperature of -10℃ and a damage size of 7mm is input into the constructed Lamb wave temperature compensation network, and the output is the temperature-compensated Lamb wave signal, which is the reconstructed signal. Simultaneously, the actual target signal corresponding to a temperature of 20℃ and a damage size of 7mm is also input as a control.

[0056] Figures 5a to 5f In the figure, the three curves represent the original input signal, the target real signal, and the reconstructed signal, respectively. The target real signal and the reconstructed signal are very close, which proves that the temperature compensation model of the present invention can achieve temperature compensation.

[0057] The damage detection network is input with both temperature-compensated and uncompensated Lamb wave signals to predict the damage size.

[0058] The root mean square error (RMSE) is used to evaluate the temperature compensation performance of the temperature compensation network method, and various error metrics are used to evaluate the damage diagnosis performance. MAE represents the mean absolute error between the predicted damage size and the actual measured damage size, and its formula is as follows:

[0059] RMSE is the square root of the mean square of the error between the actual measured damage size and the predicted damage size.

[0060] STD can reflect the degree of dispersion among individuals in a dataset, and its calculation formula is as follows:

[0061] in, Indicates the predicted damage size; This indicates the actual measured damage size; m represents the number of samples.

[0062] The three test indicators are shown in Table 2. The root mean square error of the temperature-compensated Lamb wave signal output by the temperature-compensated network model constructed in this invention for damage identification of the test component is less than 0.5 mm, while the root mean square error of the uncompensated Lamb wave signal directly input into the damage detection network for identification is greater than 0.5 mm. This indicates that the damage identification method of this invention has acceptable performance in predicting damage size.

[0063] Table 2: Model Performance

[0064] To more intuitively determine the data fluctuations, the error in predicted damage size and interquartile range at all sample points are displayed. Figure 6a and Figure 6bThe box plot shown has a red line in the middle representing the median prediction error, and a blue line representing the average number of prediction errors. The top and bottom of the boxes represent the upper and lower quartiles of the data. Values ​​between the upper and lower quartiles are described as the middle 50%. Overall, the uncompensated diagnostic method predicts damage size and results in a significantly higher quartile range of prediction errors than the method proposed in this invention.

[0065] Example 4 like Figure 8 As shown, this invention discloses a damage identification system based on Lamb wave temperature compensation, comprising the following processes: The data acquisition module is used to acquire the Lamb wave signal and actual ambient temperature of the component under test; The temperature compensation module is used to input the Lamb wave signal and the actual ambient temperature into the pre-built Lamb wave temperature compensation network model and output the temperature-compensated Lamb wave signal. The damage identification module is used to identify damage to the component under test based on the temperature-compensated Lamb wave signal and obtain the damage identification result.

[0066] Example 5 This invention also discloses a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of the damage identification method based on Lamb wave temperature compensation. The memory may include main memory, such as high-speed random access memory, or it may also include non-volatile memory, such as at least one disk storage device. The processor, network interface, and memory are interconnected via an internal bus, which may be an industry-standard architecture bus, a peripheral component interconnection standard bus, an extended industry-standard architecture bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. The memory stores the program; specifically, the program may include program code, which includes computer operation instructions. The memory may include main memory and non-volatile memory, and provides instructions and data to the processor.

[0067] Example 6 This invention also discloses a computer-readable storage medium storing a computer program. When executed by a processor, the computer program implements the steps of the damage identification method based on Lamb wave temperature compensation. Specifically, the computer-readable storage medium includes, but is not limited to, volatile memory and / or non-volatile memory. The volatile memory may include random access memory and / or cache memory, etc. The non-volatile memory may include read-only memory, hard disk, flash memory, optical disk, magnetic disk, etc.

[0068] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, optical storage, etc.) containing computer-usable program code.

[0069] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0070] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0071] These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable apparatus for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0072] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A damage identification method based on Lamb wave temperature compensation, characterized in that, The process includes the following: Acquire the Lamb wave signal and actual ambient temperature of the component under test; The Lamb wave signal and the actual ambient temperature are input into a pre-built Lamb wave temperature compensation network model, and the temperature-compensated Lamb wave signal is output. Damage identification is performed on the component under test based on the temperature-compensated Lamb wave signal, and the damage identification results are obtained. The construction process of the pre-built Lamb wave temperature compensation network model is as follows: Cracks are fabricated on the target structure of the component to be tested according to the preset crack size; A sensor is installed at the target structure to acquire Lamb wave signals of different damage sizes on the component under test at multiple different ambient temperatures as raw data; the multiple different ambient temperatures include room temperature and very low temperature. The original data is truncated to obtain the basic symmetrical modes of the Lamb wave signal, and then normalized to serve as sample data. The sample data are labeled with temperature based on the ambient temperature to obtain a sample set; the sample set is then divided into a training set and a test set. A Lamb wave temperature compensation network is constructed based on a conditional generative adversarial network; the Lamb wave temperature compensation network includes a generator and a discriminator. The training set is input into the Lamb wave temperature compensation network for adversarial training until the generator and discriminator reach Nash equilibrium. The test set is input into the trained Lamb wave temperature compensation network, which outputs the temperature-compensated Lamb wave signal until the test is passed, thus obtaining the constructed Lamb wave temperature compensation network model.

2. The damage identification method based on Lamb wave temperature compensation according to claim 1, characterized in that, The test set is input into the trained Lamb wave temperature compensation network, which outputs the temperature-compensated Lamb wave signal. This process continues until the test is passed, resulting in the constructed Lamb wave temperature compensation network model. The test set is input into the trained Lamb wave temperature compensation network, which outputs the temperature-compensated Lamb wave signal. The temperature-compensated Lamb wave signal is used to identify the target structural cracks in the component under test and output the predicted damage size. The error between the predicted damage size and the preset crack size is calculated until the error between the predicted damage size and the preset crack size meets the requirements. Then the Lamb wave temperature compensation network model is completed.

3. The damage identification method based on Lamb wave temperature compensation according to claim 1, characterized in that, The process of inputting the training set into the Lamb wave temperature compensation network for adversarial training is as follows: Train the discriminator: The original input signal and temperature label are input into the generator, which generates a reconstructed signal. The reconstructed signal and the target real signal are input into the discriminator, and the output result is fed back to the discriminator through the loss function to complete the update of the discriminator parameters; during this process, the generator parameters are frozen. Train the generator: With the parameters of the discriminator fixed, the original input signal and temperature label are input into the generator, and the generator generates the reconstructed signal. The reconstructed signal and the target real signal are input into the discriminator, and similarity is evaluated through a loss function. If the similarity evaluation meets the preset requirements, training stops; if the similarity evaluation does not meet the preset requirements, the similarity loss is backpropagated to the generator, and the generator parameters are iteratively updated until the similarity evaluation meets the preset requirements, at which point the generator training is complete. The generator and discriminator are trained alternately until Nash equilibrium is reached; The original input signal is the Lamb wave signal acquired at a very low ambient temperature, and the target real signal is the Lamb wave signal acquired at room temperature.

4. The damage identification method based on Lamb wave temperature compensation according to claim 1, characterized in that, The damage identification of the component under test based on the temperature-compensated Lamb wave signal is specifically as follows: The temperature-compensated Lamb wave signal is input into the damage detection network, and the predicted damage result is output.

5. The damage identification method based on Lamb wave temperature compensation according to claim 4, characterized in that, The damage detection network employs a convolutional neural network.

6. A damage identification system based on Lamb wave temperature compensation, characterized in that, include: The data acquisition module is used to acquire the Lamb wave signal and actual ambient temperature of the component under test; The temperature compensation module is used to input the Lamb wave signal and the actual ambient temperature into the pre-built Lamb wave temperature compensation network model and output the temperature-compensated Lamb wave signal. The damage identification module is used to identify damage to the component under test based on the temperature-compensated Lamb wave signal and obtain the damage identification result. The construction process of the pre-built Lamb wave temperature compensation network model is as follows: Cracks are fabricated on the target structure of the component to be tested according to the preset crack size; A sensor is installed at the target structure to acquire Lamb wave signals of different damage sizes on the component under test at multiple different ambient temperatures as raw data; the multiple different ambient temperatures include room temperature and very low temperature. The original data is truncated to obtain the basic symmetrical modes of the Lamb wave signal, and then normalized to serve as sample data. The sample data are labeled with temperature based on the ambient temperature to obtain a sample set; the sample set is then divided into a training set and a test set. A Lamb wave temperature compensation network is constructed based on a conditional generative adversarial network; the Lamb wave temperature compensation network includes a generator and a discriminator. The training set is input into the Lamb wave temperature compensation network for adversarial training until the generator and discriminator reach Nash equilibrium. The test set is input into the trained Lamb wave temperature compensation network, which outputs the temperature-compensated Lamb wave signal until the test is passed, thus obtaining the constructed Lamb wave temperature compensation network model.

7. A computer device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the damage identification method based on Lamb wave temperature compensation as described in any one of claims 1 to 5.

8. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the damage identification method based on Lamb wave temperature compensation as described in any one of claims 1 to 5.

9. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instruction is executed by the processor, it implements the steps of the damage identification method based on Lamb wave temperature compensation as described in any one of claims 1 to 5.

Images