An ultrasonic lamb wave dispersion compensation method and device based on convolution auto-encoding

Abstract

The application discloses an ultrasonic Lamb wave dispersion compensation method and device based on convolutional auto-encoding, which comprises the following steps: S1, constructing a convolutional auto-encoding network model applied to compensation of an ultrasonic Lamb wave dispersion signal; S2, constructing a training set data according to the time of flight of each wave packet of the ultrasonic Lamb wave dispersion signal of a multimodal multi-wave packet at different distances, and training the convolutional auto-encoding network model by using the training set data; S3, generating an ultrasonic Lamb wave test data set, testing the trained convolutional auto-encoding network model, and outputting a test result; and S4, reconstructing an ultrasonic Lamb wave signal according to the test result. The application realizes synchronous compensation of the ultrasonic Lamb wave dispersion signal of a multimodal multi-wave packet, captures the time of flight information of the dispersion signal wave packet under the condition of a multimodal multi-wave packet, can separate the superimposed wave packet, and realizes effective dispersion compensation of the ultrasonic Lamb wave signal in combination with an excitation signal waveform.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of nondestructive testing. More particularly, it relates to an ultrasonic Lamb wave dispersion compensation method and device based on convolutional auto-encoding. BACKGROUND

[0002] One of the key features of ultrasonic Lamb waves is dispersion, in which the phase velocity and group velocity depend on the product of the excitation frequency and the structure thickness. The nonlinear relationship between the wave number and the frequency leads to the diffusion of wave packets and the distortion of waveforms. In addition, when there is damage in the structure, the collected signal is the superposition of direct waves, boundary reflected waves and damage reflected waves of multiple modal components, making the signal interpretation more complex. The wave packets of dispersion are easily overlapped in the time domain, which reduces the time resolution of ultrasonic Lamb waves and further affects damage positioning and imaging results.

[0003] Current dispersion compensation is mainly achieved through traditional signal analysis and processing. One simple method is to use a narrow-band comb signal excitation under appropriate parameter settings. When the bandwidth is reduced, the dispersion effect will be weakened, and the time resolution will also be reduced. Another widely used dispersion compensation method is the time-distance mapping method, which maps the single-mode ultrasonic Lamb wave signal from the time domain to the distance domain, but the waveform of the compensated wave packet is still distorted and difficult to recover to the original excitation waveform state. However, most of the current dispersion compensation methods are only applicable to single-mode cases. If there are multiple modes in the signal, the non-target mode of the compensated ultrasonic Lamb wave will be more severely deformed, the waveform of the compensated wave packet will still be distorted, and it will be difficult to recover to the original excitation waveform state, and the mixed wave packet will be difficult to separate. In addition, when the provided prior dispersion curve data is not accurate enough, the results of the traditional method will be severely affected.

[0004] Therefore, it is necessary to provide an ultrasonic Lamb wave dispersion compensation method and device based on convolutional auto-encoding. SUMMARY

[0005] The purpose of the present application is to provide an ultrasonic Lamb wave dispersion compensation method and device based on convolutional auto-encoding, which realizes the synchronous dispersion compensation of multi-modal ultrasonic Lamb wave signals and separates the mixed wave packet, and has better performance in the case of inaccurate prior dispersion data, providing a new idea for solving ultrasonic Lamb wave signal processing tasks with deep learning technology.

[0006] To achieve the above purpose, the present application adopts the following technical solutions:

[0007] The present application provides an ultrasonic Lamb wave dispersion compensation method based on convolutional auto-encoding in the first aspect, and the method steps are:

[0008] S1, a convolutional auto-encoding network model applied to compensate ultrasonic Lamb wave dispersion signals is constructed;

[0009] S2, constructing a training set data according to the time of flight of each wave packet of the ultrasonic lamb wave dispersion signal of the multi-modal multi-wave packet under different distances, and training the convolutional auto-encoding network model by using the training set data;

[0010] S3, generating an ultrasonic lamb wave test data set, testing the trained convolutional auto-encoding network model and outputting a test result;

[0011] S4, reconstructing an ultrasonic lamb wave signal according to the test result.

[0012] Optionally, the S2 further comprises:

[0013] S21, simulating an excitation signal and obtaining an ultrasonic lamb wave dispersion signal of multi-modal multi-wave packets under different propagation distances;

[0014] S22, calculating the time of flight of each wave packet in the ultrasonic lamb wave dispersion signal according to different propagation distances and group velocities of each modal lamb wave, and generating a time sequence, the time sequence being a time sequence label corresponding to the ultrasonic lamb wave dispersion signal y(t) and constituting a training data set.

[0015] Optionally, the S22 further comprises calculating the time of flight of each wave packet in the ultrasonic lamb wave dispersion signal according to a pre-set propagation distance x S1 of the S0 modal wave packet and a propagation distance of the A0 modal wave packet.

[0016]

[0017] wherein t is time, δ(·) is an impulse function, is the group velocity of the S0 modal, is the group velocity of the A0 modal, and f c is the center frequency of the excitation signal.

[0018] Optionally, the loss function of the convolutional auto-encoding network model is a mean square error loss function L MSE .

[0019]

[0020] wherein y ToF is a time sequence result.

[0021] Optionally, the convolutional auto-encoding network model comprises a convolution module and a deconvolution module, wherein the convolution module comprises a convolution layer, a batch normalization layer and an activation function ReLU function, and is used for encoding the convolutional auto-encoding network model; the deconvolution module comprises a deconvolution layer, a batch normalization layer and an activation function ReLU function, and is used for up-sampling decoding of the convolutional auto-encoding network model.

[0022] Optionally, the test set data comprises ultrasonic Lamb wave dispersion signals excited and collected on the surface of a metal aluminum plate, and the ultrasonic Lamb wave dispersion signals are collected through paths with different propagation distances and are preprocessed.

[0023] Optionally, the S4 further comprises:

[0024] The test result is a time series result output by the convolutional auto-encoding network model, and the ultrasonic Lamb wave signal collected in the S2 is reconstructed and dispersion compensated according to the test result, and the compensated ultrasonic Lamb wave dispersion signal is output.

[0025] The second aspect of the application provides an ultrasonic Lamb wave dispersion compensation device based on convolutional auto-encoding, which comprises,

[0026] A model construction unit is configured to construct a convolutional auto-encoding network model applied to compensation of ultrasonic Lamb wave dispersion signals.

[0027] A training unit is configured to construct training set data according to the time of flight of each wave packet of the ultrasonic Lamb wave dispersion signal of the multi-modal multi-wave packet at different distances, and train the convolutional auto-encoding network model by using the training set data.

[0028] A test unit is configured to generate an ultrasonic Lamb wave test data set, test the trained convolutional auto-encoding network model and output a test result.

[0029] A signal reconstruction unit is configured to reconstruct a compensated ultrasonic Lamb wave dispersion signal according to the test result.

[0030] The third aspect of the application provides a computer device, which comprises a memory, a processor and a computer program stored in the memory and executable on the processor, and the processor implements the method provided in the first aspect of the application when executing the program.

[0031] The fourth aspect of the application provides a computer readable storage medium, which stores a computer program, and the program is executed by a processor to implement the method provided in the first aspect of the application.

[0032] The beneficial effects of the application are as follows:

[0033] The ultrasonic Lamb wave dispersion compensation method and device based on convolutional auto-encoding provided by the application can realize effective compensation of the ultrasonic Lamb wave dispersion signal, and the wave packet waveform of the compensated signal can be consistent with the excitation signal waveform. Compared with the traditional dispersion compensation signal processing, the application can realize synchronous compensation of the multi-modal multi-wave packet ultrasonic Lamb wave dispersion signal, and the time sequence result output by the convolutional auto-encoding network model can correspond to the given time sequence label, which indicates that the network can capture the dispersion signal wave packet time of flight information under the condition of multi-modal and multi-wave packet, and can separate the superimposed wave packet, and then realize effective dispersion compensation of the ultrasonic Lamb wave signal in combination with the excitation signal waveform, thereby providing a new idea for solving the ultrasonic Lamb wave signal processing task by using the deep learning technology. BRIEF DESCRIPTION OF DRAWINGS

[0034] The specific embodiments of the application will be further described in detail below with reference to the accompanying drawings.

[0035] Figure 1 A flowchart of the ultrasonic Lamb wave dispersion compensation method based on convolutional auto-encoding of the application is shown;

[0036] Figure 2 A schematic diagram of a convolutional auto-encoding network model is shown;

[0037] Figure 3 A schematic diagram of the piezoelectric ceramic sensor arrangement in the embodiment is shown;

[0038] Figure 4 The ultrasonic Lamb wave signal and the compensated signal result received by the sensor 1 in the embodiment are shown;

[0039] Figure 5 The ultrasonic Lamb wave signal and the compensated signal result received by the sensor 2 in the embodiment are shown;

[0040] Figure 6 The ultrasonic Lamb wave signal and the compensated signal result received by the sensor 3 in the embodiment are shown;

[0041] Figure 7 A schematic diagram of the computer device structure of the embodiment of the application is shown. DETAILED DESCRIPTION

[0042] In order to more clearly illustrate the application, the application will be further described below in combination with the preferred embodiments and the accompanying drawings. Similar components are denoted by the same reference numerals in the drawings. Those skilled in the art should understand that the specific description below is illustrative rather than limiting, and should not limit the protection scope of the application.

[0043] Embodiment 1

[0044] Embodiment 1 provides an ultrasonic Lamb wave dispersion compensation method based on convolutional auto-encoding, and the method steps are as follows:

[0045] S1, constructing a convolutional auto-encoding network model applied to compensate ultrasonic Lamb wave dispersion signals;

[0046] S2, constructing a training set data according to the time of flight of each wave packet of the ultrasonic Lamb wave dispersion signals of the multi-modal multi-wave packet at different distances, and training the convolutional auto-encoding network model by using the training set data;

[0047] S3, generating an ultrasonic Lamb wave test data set, testing the trained convolutional auto-encoding network model, and outputting a test result;

[0048] S4, reconstructing an ultrasonic Lamb wave signal according to the test result.

[0049] In a specific example, according to the prior dispersion curve information of the ultrasonic Lamb wave, given an excitation signal f(t), the excitation signal is simulated to obtain the ultrasonic Lamb wave dispersion signals y(t) of the multi-modal multi-wave packet at different propagation distances.

[0050] Suppose that the sensor only excites one Lamb wave mode and ignores the change of amplitude, the dispersion is caused by the nonlinear characteristics of the wave number, the duration of the response waveform increases, and the amplitude decreases. Specifically, when the propagation distance of the ultrasonic Lamb wave is x0, the response signal y0(t) collected by the sensor can be represented as:

[0051]

[0052] where t is time, ω is angular frequency, F(ω) is the Fourier transform of the excitation signal f(t), and k=k(ω) is the wave number of a certain mode of the ultrasonic Lamb wave. The excitation signal in this embodiment is a 3-period narrow-band comb signal with a center frequency of 200 kHz. Since the S0 mode and the A0 mode generally exist in the ultrasonic Lamb wave signal, the S0 and A0 modes refer to two propagation forms of ultrasonic guided waves, which are 0-order symmetric mode and 0-order anti-symmetric mode, respectively. Therefore, the dispersion signals of two wave packets under two modes in this embodiment are simulated by using the following formula:

[0053]

[0054] where k S is the prior wave number information of the S0 mode, k A is the prior wave number information of the A0 mode, x S1 is the propagation distance of the S0 mode wave packet, x A1 is the propagation distance of the A0 mode wave packet, where x S1 ∈[0.4,2.4], x A1∈[0.4,1.2], randomly take 1000 times under two wave packet conditions (1 S0 wave packet and 1 A0 wave packet), and randomly take 1000 times under four wave packet conditions (2 S0 wave packets and 2 A0 wave packets), thereby forming 2000 dispersion signal samples as training set data.

[0055] In combination with different propagation distances and Lamb wave modal group velocities, the time of flight (ToF) of each wave packet in the ultrasonic Lamb wave dispersion signal is calculated and expressed as a pulse function, respectively, to generate a time sequence containing flight time information as the label of the corresponding dispersion signal y(t), thereby forming a training data set;

[0056] When the Lamb wave signal is non-dispersive, the wave number k becomes a linear function of the frequency k', in which case the phase velocity cp(ω) and the group velocity cg(ω) of the mode satisfy the following formula:

[0057]

[0058] x is the propagation distance.

[0059] Therefore, when the propagation distance is x i (i = 1, 2, 3 … N) the dispersion compensation signal y i_c (t) can be expressed as:

[0060]

[0061] where δ(t-t i ) is a pulse function. Further, the corresponding dispersion compensation signal can be expressed as:

[0062]

[0063] where t is the time, t S1 is the propagation time of the S0 modal wave packet, t A1 is the propagation time of the A0 modal wave packet, is the group velocity of the S0 mode, is the group velocity of the A0 mode, f c is the center frequency of the excitation signal, which is set to 200 kHz, according to the propagation distance x S1 and x A1 , the time sequence containing flight time information is calculated and used as the time sequence label of the corresponding ultrasonic Lamb wave dispersion signal

[0064]

[0065] It is input into the convolutional auto-encoding network model as the label of the training set data.

[0066] The acquired simulation data and labels are input into the convolutional auto-encoding network model as a training set for training.

[0067] The schematic diagram of the convolutional auto-encoding network model is shown in FIG. 2. Figure 2 The convolutional auto-encoding network model includes two convolutional modules and two deconvolutional modules. The convolutional module includes a convolutional layer, a batch normalization (BN) layer and an activation function ReLU function, and encodes the convolutional auto-encoding network model. Correspondingly, the deconvolutional module includes a deconvolutional layer, a BN layer and a ReLU function, and decodes the convolutional auto-encoding network model by upsampling. The kernel sizes of the convolutional layer and the deconvolutional layer are 1x10, 1x3, 1x3 and 1x10, respectively, and the numbers thereof are 4, 8, 8 and 4, respectively. The dropout operation is used before the deconvolutional module to avoid overfitting, and the ratio is set to 0.25. After feature extraction, the time series is regenerated, the mean square error is used as the loss function, the stochastic gradient descent is used as the optimizer, the learning rate is set to 0.0001, and the batch size is set to 64 for training.

[0068] The loss function of the convolutional auto-encoding network model adopts a mean square error loss function L MSE ,

[0069]

[0070] y ToF is the time series result.

[0071] The piezoelectric ceramic sensor is placed on the surface of the metal aluminum plate to excite and collect ultrasonic Lamb wave signals. Three paths with different propagation distances are set on the aluminum plate specimen. The excitation signal is a 3-period narrow-band comb signal with a center frequency of 200 kHz. The frequency dispersion signal is preprocessed to form an ultrasonic Lamb wave test data set. In this embodiment, an aluminum plate with a specification of 2000mmx1200mmx2mm is used as a specimen, Figure 3 is a schematic diagram of the arrangement of the piezoelectric ceramic sensor.

[0072] The acquired test data set is input into the trained convolutional auto-encoding network model as a test set to test the trained convolutional auto-encoding network model. The time of flight of each wave packet in each frequency dispersion signal is obtained, and the test result is output. The test result is the time series result y ToF output by the convolutional auto-encoding network model.

[0073] According to the test result, the ultrasonic Lamb wave signal collected in S2 is reconstructed and dispersion compensated. The excitation signal f(t) is convolved with the time series result y ToF , multiplied by the amplitude at the corresponding time of the original collected signal, and the dispersion compensated ultrasonic Lamb wave y c(t). The original collected ultrasonic Lamb wave signals at three distances of the receiving sensor and the network compensated signals are shown in Figs. Figure 4 、 Figure 5 and Figure 6 respectively. Due to the boundary effect, there are direct waves of S0 mode and A0 mode and boundary reflection waves in the signals. The results show that the true ToF can correspond to the ToF obtained by the convolutional auto-encoding network model output, and the clear dispersion compensation signal is obtained after convolution excitation.

[0074] It can be found that the convolutional auto-encoding model proposed in the technical scheme can realize the dispersion compensation of ultrasonic Lamb wave signals. The results show that the present application can complete the synchronous compensation of multi-modal ultrasonic Lamb wave dispersion signals, overcome the problem that the traditional signal processing method can only compensate single-mode signals, and the method can effectively separate the overlapping wave packets. The present application has high intelligence and is more applicable in engineering practice.

[0075] The technical scheme proposed in the present application can realize the synchronous compensation of multi-modal ultrasonic Lamb wave dispersion signals. Through analysis of the experimental results, it can be found that the time sequence obtained by the network model output can correspond to the given time sequence label, which shows that the network can capture the dispersion signal wave packet time of flight information under the conditions of multi-modal and multi-wave packet, and can separate the overlapping wave packets, and further realize the effective dispersion compensation of ultrasonic Lamb wave signals in combination with the excitation signal waveform.

[0076] Data-driven methods have received more and more attention due to their significant feature extraction capability, and many deep learning methods have been applied in nondestructive testing and structural health monitoring. Current deep learning models are mostly used to extract damage features, and then directly perform classification or regression tasks on the signals, and have not identified waveform features from the signals. One of the innovations of the present application is to use deep learning methods to solve the traditional signal processing task of Lamb wave, starting from the perspective of waveform analysis, realizing the dispersion compensation of the signal, and weakening the influence of weak prior dispersion information. In addition, the present application can overcome the limitation that the traditional dispersion compensation method is only applicable to single-mode, and can complete the synchronous dispersion compensation of multi-modal ultrasonic Lamb wave signals, and separate the overlapping wave packets in the signal. The present application has high intelligence and wide application prospect.

[0077] Embodiment 2

[0078] Embodiment 2 provides an ultrasonic Lamb wave dispersion compensation device based on convolutional auto-encoding, which comprises,

[0079] a model construction unit configured to construct a convolutional auto-encoding network model applied to compensate ultrasonic Lamb wave dispersion signals;

[0080] The training unit is configured to construct training set data according to the time of flight of each wave packet of the ultrasonic Lamb wave dispersion signal of the multi-modal multi-wave packet at different distances, and train the convolutional auto-encoding network model by using the training set data.

[0081] The testing unit is configured to generate an ultrasonic Lamb wave test data set, test the trained convolutional auto-encoding network model, and output a test result.

[0082] The signal reconstruction unit is configured to reconstruct the compensated ultrasonic Lamb wave dispersion signal according to the test result.

[0083] Embodiment 3

[0084] As shown in Figure 7 Embodiment 3 provides a computer device, and it can be understood that, Figure 7 The computer device 12 shown is merely one example and should not be taken as limiting the scope of functionality or use of embodiments of the application.

[0085] As shown in Figure 7 The computer device 12 is presented in the form of a general-purpose computing device. The components of the computer device 12 can include, but are not limited to, one or more processors or processing units 16, a system memory 28, and a bus 18 that couples various system components including the system memory 28 and the processing unit 16.

[0086] The bus 18 represents one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, a graphics acceleration bus, a processor or local bus using any of a variety of bus architectures including an industry standard architecture (ISA), micro-channel architecture (MAC), enhanced ISA (EISA), Video Electronics Standards Association (VESA) local bus, and a peripheral component interconnect (PCI) bus.

[0087] The computer device 12 typically includes a variety of computer system readable media. Such media can be any available media that is located either in or out of the computer device 12, including both volatile and non-volatile media, removable and non-removable media.

[0088] The system memory 28 can include computer system readable media in the form of volatile memory, such as random access memory (RAM) 30 and / or cache memory 32. The computer device 12 can further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, a storage system 34 can be provided for reading from and writing to non-removable, non-volatile magnetic media (e.g., a "hard drive"). Figure 7 Not shown is a generally-omitted "hard disk drive", typically providing non-volatile memory in the form of an Figure 7A disk drive, a floppy disk drive, a CD-ROM drive, a DVD-ROM drive, or other removable media drive, or any combination thereof, can be provided in the computer device 12 in addition to or instead of the disk drive illustrated in FIG. 1. In such instances, each drive can be connected to the system bus 18 by one or more data media interfaces. The memory 28 can include at least one program product having a set (e.g., at least one) of program modules that are configured to carry out the functions of embodiments of the application.

[0089] The program / utility 40, having a set (at least one) of program modules 42, can be stored in memory 28 by way of example, and not limitation, as well as an operating system, one or more application programs, other program modules, and program data, each or some combination thereof, may

[0090] The computer device 12 can also communicate with one or more external devices 14 such as a keyboard, a pointing device, a display 24, etc.; one or more devices that enable a user to interact with the computer device 12; and / or one or more devices that enable the computer device 12 to communicate with one or more other computer devices. Such communication can be via an input / output (I / O) interface 22. Still yet, the computer device 12 can communicate with one or more networks, such as a local area network (LAN), a general wide area network (WAN), and / or a public network, such as the Internet, via a network adapter 20. As Figure 7 illustrated, the network adapter 20 communicates with the other components of the computer device 12 via the bus 18. It should be appreciated that although the network adapter 20 is illustrated as a single component in FIG. 1, the network adapter 20 can comprise two or more components that work together to facilitate communications with one or more other computer devices. ​ It should be appreciated that the software modules described herein can be stored in the memory 28 and loaded into the processor 16 when it is desired to execute the software modules. In addition, the software modules can be loaded into the memory 28 as virtual machines.

[0091] The processor unit 16 performs various function applications and data processing by running programs stored in the system memory 28, such as implementing the method provided by embodiment 1.

[0092] Embodiment 4

[0093] Embodiment 4 provides a computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements the method described in embodiment 1.

[0094] In practical application, the computer-readable storage medium can adopt any combination of one or more computer-readable media. The computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium may, for example, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or apparatus, or any combination thereof. More specific examples (a non-exhaustive list) of the computer-readable storage medium include an electrical connection having one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this embodiment, the computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in connection with an instruction execution system, apparatus or device.

[0095] The computer-readable signal medium can include a computer-readable program code in a baseband or propagated as a carrier wave in a propagation medium. Such a propagated signal can take a wide variety of forms, including but not limited to, electro-magnetic, optical, or any suitable combination thereof. Computer-readable signal media can also be any computer-readable medium that is not a storage medium and that can communicate or propagate program code instructions; such computer-readable signal media can include, but are not limited to, an electrical connection (electrical communication), a fiber optic cable, and / or a physical transmission such as an electromagnetic transmission or a radio frequency transmission.

[0096] Program code embodied on a computer-readable medium can be transmitted using any appropriate medium, including but not limited to wireless, wire line, optical fiber cable, RF, etc., or any suitable combination of the above.

[0097] Computer program code for carrying out operations of the present application can be written in any combination of one or more programming languages, including an object oriented programming language such as Java, Smalltalk, C++ or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code can execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection can be made to an external computer (for example, through the Internet using an Internet Service Provider). In an embodiment, multiple data storage devices can be used to store the program code.

[0098] Obviously, the above embodiments of the present application are merely exemplary and are not intended to limit the embodiments of the present application. Based on the above description, other different forms of changes or variations can be made by those skilled in the art, and it is impossible to enumerate all the embodiments here. Any obvious changes or variations derived from the technical solutions of the present application are still within the protection scope of the present application.

Claims

1. An ultrasonic Lamb wave dispersion compensation method based on convolutional auto-encoding, characterized by, The method comprises, S1, constructing a convolutional auto-encoding network model applied to compensate an ultrasonic Lamb wave dispersion signal; S2, constructing a training set data according to a time of flight of each wave packet of the ultrasonic Lamb wave dispersion signal of a multi-modal multi-wave packet under different distances, and training the convolutional auto-encoding network model by using the training set data; S3, generating an ultrasonic Lamb wave test data set, testing the trained convolutional auto-encoding network model, and outputting a test result; S4, reconstructing an ultrasonic Lamb wave signal according to the test result; The S2 further comprises: S21, simulating an excitation signal and obtaining an ultrasonic Lamb wave dispersion signal of a multi-modal multi-wave packet under different propagation distances; S22. Based on different propagation distances and the velocities of each mode group of the Lamb wave, calculate the flight time of each wave packet in the ultrasonic Lamb wave dispersion signal and generate a time series. The time series serves as the corresponding ultrasonic Lamb wave dispersion signal. The time series labels constitute the training dataset; S22 further comprises calculating a propagation distance of the S0 modal wave packet according to a pre-set and a propagation distance of the A0 modal wave packet a time series label containing time-of-flight information , ， where t is time, is a pulse function, is the group velocity of the S0 mode, is the group velocity of the A0 mode, is the center frequency of the excitation signal; The loss function of the convolutional auto-encoding network model is a mean square error loss function L MSE , ， Wherein, yToF is a time series result.

2. The ultrasonic Lamb wave dispersion compensation method of claim 1, wherein, The convolutional auto-encoding network model comprises a convolutional module and a deconvolutional module, wherein, The convolutional module comprises a convolutional layer, a batch normalization layer, and an activation function ReLU function, and is used for encoding the convolutional auto-encoding network model; The deconvolutional module comprises a deconvolutional layer, a batch normalization layer, and an activation function ReLU function, and is used for up-sampling decoding of the convolutional auto-encoding network model.

3. The ultrasonic Lamb wave dispersion compensation method of claim 1, wherein, The test data set comprises an ultrasonic Lamb wave dispersion signal excited and collected on a surface of a metal aluminum plate, and the ultrasonic Lamb wave dispersion signal is obtained by setting different propagation distance paths and preprocessing.

4. The ultrasonic Lamb wave dispersion compensation method of claim 1, wherein, The S4 further comprises: The test result is a time series result output by the convolutional auto-encoding network model, and the ultrasonic Lamb wave signal collected in the S2 is reconstructed and compensated for dispersion, and an ultrasonic Lamb wave dispersion signal after compensation is output.

5. An ultrasonic Lamb wave dispersion compensation device based on convolutional auto-encoding for implementing the compensation method according to any one of claims 1-4, characterized in that, Comprise, A model construction unit is configured to construct a convolutional auto-encoding network model applied to compensate an ultrasonic Lamb wave dispersion signal; A training unit is configured to construct a training set data according to a time of flight of each wave packet of the ultrasonic Lamb wave dispersion signal of a multi-modal multi-wave packet under different distances, and train the convolutional auto-encoding network model by using the training set data; A test unit is configured to generate an ultrasonic Lamb wave test data set, test the trained convolutional auto-encoding network model, and output a test result; A signal reconstruction unit is configured to reconstruct an ultrasonic Lamb wave dispersion signal after compensation according to the test result.

6. A computer device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor executes the program to realize the method in any one of claims 1-4.

7. A computer-readable storage medium having stored thereon a computer program, characterized in that The program is executed by the processor to realize the method in any one of claims 1-4.

Images