Method, device, equipment and storage medium for defect detection of air bubble lightweight soil

Abstract

The present application discloses a method, device, equipment and storage medium for defect detection of bubble lightweight soil, wherein the method for defect detection of bubble lightweight soil comprises: obtaining an air-coupled impact echo signal; determining a time-frequency signal image based on the air-coupled impact echo signal; inputting the time-frequency signal image into a preset defect detection model, and performing defect recognition processing on the time-frequency signal image based on the defect detection model to obtain defect data, wherein the defect detection model is obtained by training the time-frequency signal image samples and the Transformer model block. In the method of using air-coupled impact echoes to detect internal defects of concrete structures, a time-frequency signal image is generated from the air-coupled impact echo signal, and then the defect detection model composed of the Transformer model blocks is used to perform defect analysis and recognition on the time-frequency signal image, thereby improving the accuracy of defect information detection.

Description

Technical Field

[0001] The present application relates to the field of defect detection, and in particular to a method, device, equipment and storage medium for defect detection of bubble lightweight soil. Background Art

[0002] The air-coupled impact echo method uses air as the stress wave coupling medium, and does not require the sensor to be in direct contact with the surface of the object being measured. This method has begun to be applied to the internal defect detection of concrete components. However, since the internal non-defect pores in concrete will interfere with its acoustic field characteristics and the resonance mode of the impact echo, the accuracy of defect information detection is low. Summary of the invention

[0003] The main purpose of this application is to provide a method, device, equipment and storage medium for defect detection of bubble lightweight soil, aiming to solve the technical problem of low accuracy of defect information detection in the prior art.

[0004] To achieve the above objectives, the present application provides a method for detecting defects in light-weight soil with bubbles, the method comprising:

[0005] Acquire air-coupled impact echo signal;

[0006] Determining a time-frequency signal image based on the air-coupled impact echo signal;

[0007] The time-frequency signal image is input into a preset defect detection model, and based on the defect detection model, defect recognition processing is performed on the time-frequency signal image to obtain defect data, wherein the defect detection model is obtained by training the time-frequency signal image samples and the Transformer model block.

[0008] Optionally, before the step of acquiring the air-coupled impact echo signal, the method includes:

[0009] Obtaining a Transformer model block, a time-frequency signal image sample, and a defect data label of the time-frequency signal image sample;

[0010] Based on the Transformer model block, the time-frequency signal image samples and the defect data labels of the time-frequency signal image samples, the model to be trained is trained by a sequence-to-sequence training method to obtain the defect detection model that meets the accuracy conditions.

[0011] Optionally, the step of training the model to be trained by a sequence-to-sequence training method based on the Transformer model block, the time-frequency signal image samples and the defect data labels of the time-frequency signal image samples to obtain the defect detection model that meets the accuracy condition includes:

[0012] Based on the Transformer model block, normalizing the time-frequency signal image samples to obtain normalized time-frequency signal image samples;

[0013] Inputting the normalized time-frequency signal image samples into the model to be trained to obtain predicted defect data;

[0014] Based on the predicted defect data and the defect data labels of the time-frequency signal image samples, the model to be trained is trained by the sequence-to-sequence training method to obtain the defect detection model.

[0015] Optionally, the step of performing defect recognition processing on the time-frequency signal image based on the defect detection model to obtain defect data includes:

[0016] Based on the defect detection model, extracting features from the time-frequency signal image to obtain feature results;

[0017] Based on the characterization results, the defect data is determined.

[0018] Optionally, the step of extracting features from the time-frequency signal image based on the defect detection model to obtain feature results includes:

[0019] Get the target query vector;

[0020] Based on the Transformer model block in the defect detection model, the time-frequency signal image is serialized to obtain each serialization feature;

[0021] Each of the serialized features is fused to obtain a global feature;

[0022] The feature result is obtained based on the global feature and the target query vector.

[0023] Optionally, the step of determining the defect data based on the feature result includes:

[0024] Nonlinear mapping is performed on the feature sequence to obtain the defect data.

[0025] Optionally, the step of determining a time-frequency signal image based on the air-coupled impact echo signal includes:

[0026] Based on the air-coupled impulse echo signal, determining a time domain signal and a frequency domain signal of the air-coupled impulse echo signal;

[0027] The waveform images of the time domain signal and the frequency domain signal are combined to obtain the time-frequency signal image.

[0028] The present application also provides a bubble light soil defect detection device, the bubble light soil defect detection device comprising:

[0029] An acquisition module, used for acquiring an air-coupled impact echo signal;

[0030] An image determination module, used to determine a time-frequency signal image based on the air-coupled impact echo signal;

[0031] The recognition module is used to input the time-frequency signal image into a preset defect detection model, and based on the defect detection model, perform defect recognition processing on the time-frequency signal image to obtain defect data, wherein the defect detection model is obtained by training the time-frequency signal image samples and the Transformer model block.

[0032] The present application also provides a bubble light soil defect detection device, the bubble light soil defect detection device comprising: a memory, a processor, and a program stored in the memory for implementing the bubble light soil defect detection method.

[0033] The memory is used to store a program for implementing a method for detecting defects in air bubble lightweight soil;

[0034] The processor is used to execute a program for implementing the bubble lightweight soil defect detection method to implement the steps of the bubble lightweight soil defect detection method.

[0035] The present application also provides a storage medium, on which is stored a program for implementing the bubble lightweight soil defect detection method, and the program for implementing the bubble lightweight soil defect detection method is executed by a processor to implement the steps of the bubble lightweight soil defect detection method.

[0036] The present application provides a method, device, equipment and storage medium for defect detection of bubble lightweight soil. Compared with the low accuracy of defect information detection in the prior art, in the present application, an air-coupled impact echo signal is obtained; based on the air-coupled impact echo signal, a time-frequency signal image is determined; the time-frequency signal image is input into a preset defect detection model, and based on the defect detection model, the time-frequency signal image is subjected to defect recognition processing to obtain defect data, wherein the defect detection model is obtained by training the time-frequency signal image samples and the Transformer model block. That is, in the present application, in the method of detecting internal defects of concrete structures using impact echoes, a time-frequency signal image is generated by an air-coupled impact echo signal, and then a defect detection model composed of Transformer model blocks is used to perform defect analysis and identification on the time-frequency signal image, thereby improving the accuracy of defect information detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] The drawings herein are incorporated into the specification and constitute a part of the specification, illustrate embodiments consistent with the present application, and are used together with the specification to explain the principles of the present application. In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings required for use in the embodiments or the prior art description will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0038] Figure 1 It is a schematic diagram of the device structure of the hardware operating environment involved in the embodiment of the present application;

[0039] Figure 2 This is a flow chart of the first embodiment of the bubble lightweight soil defect detection method of the present application.

[0040] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0041] It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0042] like Figure 1 As shown, Figure 1 It is a schematic diagram of the terminal structure of the hardware operating environment involved in the embodiment of the present application.

[0043] The terminal in the embodiment of the present application can be a PC, or it can be a smart phone, a tablet computer, an e-book reader, an MP3 (Moving Picture Experts Group Audio Layer III) player, an MP4 (Moving Picture Experts Group Audio Layer IV) player, a portable computer, or other portable terminal devices with display function.

[0044] like Figure 1As shown, the terminal may include: a processor 1001, such as a CPU, a network interface 1004, a user interface 1003, a memory 1005, and a communication bus 1002. Among them, the communication bus 1002 is used to realize the connection and communication between these components. The user interface 1003 may include a display screen (Display), an input unit such as a keyboard (Keyboard), and the optional user interface 1003 may also include a standard wired interface and a wireless interface. The network interface 1004 may optionally include a standard wired interface and a wireless interface (such as a WI-FI interface). The memory 1005 may be a high-speed RAM memory, or a stable memory (non-volatile memory), such as a disk memory. The memory 1005 may also be a storage device independent of the aforementioned processor 1001.

[0045] Optionally, the terminal may also include a camera, an RF (Radio Frequency) circuit, a sensor, an audio circuit, a WiFi module, and the like. Among them, sensors include light sensors, motion sensors, and other sensors. Specifically, the light sensor may include an ambient light sensor and a proximity sensor, wherein the ambient light sensor may adjust the brightness of the display screen according to the brightness of the ambient light, and the proximity sensor may turn off the display screen and / or backlight when the mobile terminal is moved to the ear. As a type of motion sensor, the gravity acceleration sensor can detect the magnitude of acceleration in each direction (generally three axes), and can detect the magnitude and direction of gravity when stationary. It can be used for applications that identify the posture of the mobile terminal (such as horizontal and vertical screen switching, related games, magnetometer posture calibration), vibration recognition related functions (such as pedometer, tapping), etc.; of course, the mobile terminal may also be equipped with other sensors such as gyroscopes, barometers, hygrometers, thermometers, infrared sensors, etc., which will not be repeated here.

[0046] Those skilled in the art will understand that Figure 1 The terminal structure shown in the figure does not constitute a limitation on the terminal, and may include more or less components than shown in the figure, or combine certain components, or arrange the components differently.

[0047] like Figure 1 As shown, the memory 1005 as a computer storage medium may include an operating device, a network communication module, a user interface module, and a bubble lightweight soil defect detection program.

[0048] exist Figure 1 In the terminal shown, the network interface 1004 is mainly used to connect to the background server and communicate data with the background server; the user interface 1003 is mainly used to connect to the client (user end) and communicate data with the client; and the processor 1001 can be used to call the bubble lightweight soil defect detection program stored in the memory 1005.

[0049] Reference Figure 2 The embodiment of the present application provides a method for detecting defects in light soil with bubbles, and the method for detecting defects in light soil with bubbles includes:

[0050] Step S100, acquiring an air-coupled impact echo signal;

[0051] Step S200, determining a time-frequency signal image based on the air-coupled impact echo signal;

[0052] Step S300, inputting the time-frequency signal image into a preset defect detection model, and performing defect recognition processing on the time-frequency signal image based on the defect detection model to obtain defect data, wherein the defect detection model is obtained by training the time-frequency signal image samples and the Transformer model block.

[0053] In this embodiment, the specific application scenario may be:

[0054] The air-coupled impact echo method is used to detect internal defects of concrete defects in planar structures. However, in air-filled lightweight concrete, the acoustic field characteristics and the resonance mode of the impact echo interfere with the information judgment of the defects, and the accuracy of defect information detection is low.

[0055] The specific steps are as follows:

[0056] The present application also provides a method for detecting defects in bubble-filled lightweight soil, the method comprising:

[0057] Step S100, acquiring an air-coupled impact echo signal;

[0058] In this embodiment, the bubble lightweight soil defect detection method is applied to a bubble lightweight soil defect detection device. First, the device obtains an air-coupled impact echo signal, wherein the air-coupled impact echo signal is a signal of a reflected stress wave. The device includes a soundproof cover, an impactor, a sound wave receiver, an acceleration sensor, and a bubble lightweight concrete component. Specifically, the device obtains the air-coupled impact echo signal in the following manner:

[0059] The impactor hits the bubble lightweight concrete component, and stress waves are generated at the striking surface and transmitted to the inside of the bubble lightweight concrete component. There are void defects in the bubble lightweight concrete component. Since the acoustic impedance value of the bubble lightweight concrete at the void defect is extremely different, the stress wave will be reflected. The soundproof cover covers the impactor to isolate the test error caused by the transmission of sound waves through the external air. The sound wave receiver and the acceleration sensor are attached directly above the soundproof cover. The device receives the reflected stress wave and records the impact air-coupled impact echo signal.

[0060] Step S200, determining a time-frequency signal image based on the air-coupled impact echo signal;

[0061] In this embodiment, the device analyzes and processes the air-coupled impact echo signal to obtain a time-frequency signal image.

[0062] Specifically, the step S200 includes the following steps S210-S220:

[0063] Step S210, determining a time domain signal and a frequency domain signal of the air-coupled impulse echo signal based on the air-coupled impulse echo signal;

[0064] In this embodiment, the device determines the time domain signal and frequency domain signal of the air-coupled shock echo signal based on the air-coupled shock echo signal by fast Fourier transform or wavelet analysis to obtain the time domain signal and frequency domain signal of the air-coupled shock echo signal.

[0065] Step S220: combining the waveform images of the time domain signal and the frequency domain signal to obtain the time-frequency signal image.

[0066] In this embodiment, the device combines the waveform images of the time domain signal and the frequency domain signal into a time-frequency signal image, wherein the horizontal axis of the image is time, the vertical axis is frequency, different block colors represent corresponding signal amplitudes, and includes waveform images of the time domain signal and the frequency domain signal.

[0067] Step S300, inputting the time-frequency signal image into a preset defect detection model, and performing defect recognition processing on the time-frequency signal image based on the defect detection model to obtain defect data, wherein the defect detection model is obtained by training the time-frequency signal image samples and the Transformer model block.

[0068] In this embodiment, the device uses a defect detection model composed of Transformer model blocks to perform defect analysis and identification on the time-frequency signal image, thereby improving the accuracy of defect information detection.

[0069] Specifically, the step S300 includes the following steps S310-S320:

[0070] Step S310, based on the defect detection model, extracting features from the time-frequency signal image to obtain feature results;

[0071] In this embodiment, the device extracts features from the time-frequency signal image based on the defect detection model to obtain feature results.

[0072] Specifically, the step S310 includes the following steps S311-S314:

[0073] Step S311, obtaining a target query vector;

[0074] Step S312, based on the Transformer model block in the defect detection model, serialize the time-frequency signal image to obtain each serialization feature;

[0075] Step S313, fusing each of the serialized features to obtain a global feature;

[0076] Step S314: obtaining the feature result based on the global feature and the target query vector.

[0077] In this embodiment, the Transformer model block in the defect detection model includes three parts: a feature extraction network Backbone, an encoder-decoder network Encoder-Decoder, and a prediction network Predictor. The encoder-decoder network uses a multi-head self-attention module Multi-Head Self Attention to extract different feature information of the input sequence by mapping the input feature sequence to multiple feature subspaces. First, the feature extraction network uses a Multi-Scale Transformer backbone network to extract feature maps containing defect information in the time-frequency signal image. Then, the encoder-decoder network uses the Pre-LN Transformer structure to first expand the feature map along the spatial dimension to obtain each serialized feature, and uses the Encoder structure in the Pre-LN Transformer to fuse the input serialized features to obtain the global features of the image. The Decoder structure obtains the feature results based on the global feature representation of the input and the target query vector.

[0078] Step S320: determining the defect data based on the feature result.

[0079] Specifically, the step S320 includes the following steps S321:

[0080] Step S321, performing nonlinear mapping on the feature sequence to obtain the defect data.

[0081] In this embodiment, the prediction network may use a multi-layer perceptron to perform nonlinear mapping on the input crack feature results to obtain a set of prediction boxes containing defect information, wherein the defect data includes data on the depth, shape and size of the defect.

[0082] Before the step S100, the step of acquiring the air-coupled impact echo signal, the method includes the following steps A100-A200:

[0083] Step A100, obtaining a Transformer model block, a time-frequency signal image sample, and a defect data label of the time-frequency signal image sample;

[0084] In this embodiment, the defect detection model is a model composed of multiple layers of Transformer blocks, the time-frequency signal image samples are image samples used for model training, and the defect data labels of the time-frequency signal image samples are labels of defect data corresponding to the time-frequency signal image samples.

[0085] Step A200, based on the Transformer model block, the time-frequency signal image samples and the defect data labels of the time-frequency signal image samples, the model to be trained is trained by a sequence-to-sequence training method to obtain the defect detection model that meets the accuracy conditions.

[0086] In this embodiment, the device trains the model to be trained by a sequence-to-sequence training method based on the Transformer model block, the time-frequency signal image sample and the defect data label of the time-frequency signal image sample to obtain the defect detection model that meets the accuracy condition.

[0087] Specifically, the step A200 includes the following steps A210-A230:

[0088] Step A210, normalizing the time-frequency signal image samples based on the Transformer model block to obtain normalized time-frequency signal image samples;

[0089] In this embodiment, the Pre-LN Transformer structure is used in the encoder-decoder stage, and layer normalization is used inside the residual block to effectively reduce the parameter gradient near the output layer and accelerate the model convergence.

[0090] Step A220, inputting the normalized time-frequency signal image sample into the model to be trained to obtain predicted defect data;

[0091] Step A230: Based on the predicted defect data and the defect data labels of the time-frequency signal image samples, the model to be trained is trained by the sequence-to-sequence training method to obtain the defect detection model.

[0092] In this embodiment, the predicted defect data and the defect data label of the time-frequency signal image sample are difference calculated to obtain an error result;

[0093] Based on the error result, determining whether the error result meets an error standard indicated by a preset error threshold range;

[0094] If the error result does not meet the error standard indicated by the preset error threshold range, return to the step of inputting the normalized time-frequency signal image sample into the model to be trained to obtain predicted defect data, and stop training until the training error result meets the error standard indicated by the preset error threshold range, that is, the model convergence is completed, and the defect detection model is obtained.

[0095] In another embodiment, an auxiliary training loss function is added to the decoder structure to improve the accuracy of predicting target categories, improve the accuracy of the above model training process, and further accelerate the model convergence speed.

[0096] The present application provides a method for defect detection of bubble lightweight soil. Compared with the low accuracy of defect information detection in the prior art, in the present application, an air-coupled impact echo signal is obtained; based on the air-coupled impact echo signal, a time-frequency signal image is determined; the time-frequency signal image is input into a preset defect detection model, and based on the defect detection model, defect recognition processing is performed on the time-frequency signal image to obtain defect data, wherein the defect detection model is obtained by training the time-frequency signal image samples and the Transformer model block. That is, in the present application, in the method of detecting internal defects of concrete structures using impact echoes, a time-frequency signal image is generated by an air-coupled impact echo signal, and then a defect detection model composed of Transformer model blocks is used to perform defect analysis and identification on the time-frequency signal image, thereby improving the accuracy of defect information detection.

[0097] The present application also provides a bubble light soil defect detection device, the bubble light soil defect detection device comprising:

[0098] An acquisition module, used for acquiring an air-coupled impact echo signal;

[0099] An image determination module, used to determine a time-frequency signal image based on the air-coupled impact echo signal;

[0100] The recognition module is used to input the time-frequency signal image into a preset defect detection model, and based on the defect detection model, perform defect recognition processing on the time-frequency signal image to obtain defect data, wherein the defect detection model is obtained by training the time-frequency signal image samples and the Transformer model block.

[0101] Optionally, the bubble lightweight soil defect detection device further includes:

[0102] A sample acquisition module, used to acquire a Transformer model block, a time-frequency signal image sample, and a defect data label of the time-frequency signal image sample;

[0103] A training module is used to train the model to be trained through a sequence-to-sequence training method based on the Transformer model block, the time-frequency signal image samples and the defect data labels of the time-frequency signal image samples, so as to obtain the defect detection model that meets the accuracy conditions.

[0104] Optionally, the training module includes:

[0105] A normalization module, used for normalizing the time-frequency signal image samples based on the Transformer model block to obtain normalized time-frequency signal image samples;

[0106] An input module, used for inputting the normalized time-frequency signal image samples into the model to be trained to obtain predicted defect data;

[0107] A model training module is used to train the model to be trained based on the predicted defect data and the defect data labels of the time-frequency signal image samples by the sequence-to-sequence training method to obtain the defect detection model.

[0108] Optionally, the identification module includes:

[0109] A feature extraction module, used to extract features from the time-frequency signal image based on the defect detection model to obtain feature results;

[0110] A defect data determination module is used to determine the defect data based on the feature result.

[0111] Optionally, the feature extraction module includes:

[0112] A vector acquisition module is used to acquire a target query vector;

[0113] A serialization module, used for serializing the time-frequency signal image based on the Transformer model block in the defect detection model to obtain each serialization feature;

[0114] A feature fusion module, used for fusing each of the serialized features to obtain a global feature;

[0115] A result determination module is used to obtain the feature result based on the global feature and the target query vector.

[0116] Optionally, the defect data determination module includes:

[0117] A mapping module is used to perform nonlinear mapping on the feature sequence to obtain the defect data.

[0118] Optionally, the image determination module includes:

[0119] A signal determination module, used to determine a time domain signal and a frequency domain signal of the air-coupled impulse echo signal based on the air-coupled impulse echo signal;

[0120] The combining module is used to combine the waveform images of the time domain signal and the frequency domain signal to obtain the time-frequency signal image.

[0121] The specific implementation of the bubble lightweight soil defect detection device of the present application is basically the same as the various embodiments of the bubble lightweight soil defect detection method mentioned above, and will not be repeated here.

[0122] Reference Figure 1 , Figure 1 It is a schematic diagram of the terminal structure of the hardware operating environment involved in the embodiment of the present application.

[0123] like Figure 1 As shown, the terminal may include: a processor 1001, such as a CPU, a network interface 1004, a user interface 1003, a memory 1005, and a communication bus 1002. Among them, the communication bus 1002 is used to realize the connection and communication between these components. The user interface 1003 may include a display screen (Display), an input unit such as a keyboard (Keyboard), and the optional user interface 1003 may also include a standard wired interface and a wireless interface. The network interface 1004 may optionally include a standard wired interface and a wireless interface (such as a WI-FI interface). The memory 1005 may be a high-speed RAM memory, or a stable memory (non-volatile memory), such as a disk memory. The memory 1005 may also be a storage device independent of the aforementioned processor 1001.

[0124] Optionally, the bubble lightweight soil defect detection device may also include a rectangular user interface, a network interface, a camera, an RF (Radio Frequency) circuit, a sensor, an audio circuit, a WiFi module, etc. The rectangular user interface may include a display screen (Display), an input submodule such as a keyboard (Keyboard), and the optional rectangular user interface may also include a standard wired interface and a wireless interface. The network interface may optionally include a standard wired interface and a wireless interface (such as a WI-FI interface).

[0125] Those skilled in the art will understand that Figure 1The structure of the bubble lightweight soil defect detection equipment shown in the figure does not constitute a limitation of the bubble lightweight soil defect detection equipment, and may include more or fewer components than shown in the figure, or a combination of certain components, or a different arrangement of components.

[0126] like Figure 1 As shown, the memory 1005 as a storage medium may include an operating system, a network communication module, and a bubble light soil defect detection program. The operating system is a program that manages and controls the hardware and software resources of the bubble light soil defect detection device, and supports the operation of the bubble light soil defect detection program and other software and / or programs. The network communication module is used to realize the communication between the components inside the memory 1005, and the communication with other hardware and software in the bubble light soil defect detection system.

[0127] exist Figure 1 In the bubble lightweight soil defect detection device shown, the processor 1001 is used to execute the bubble lightweight soil defect detection program stored in the memory 1005 to implement the steps of any of the bubble lightweight soil defect detection methods described above.

[0128] The specific implementation methods of the bubble lightweight soil defect detection equipment of the present application are basically the same as the embodiments of the bubble lightweight soil defect detection method described above, and will not be repeated here.

[0129] The present application also provides a storage medium, on which is stored a program for implementing a method for detecting defects in light soil with bubbles. The program for implementing a method for detecting defects in light soil with bubbles is executed by a processor to implement the following method for detecting defects in light soil with bubbles:

[0130] Acquire air-coupled impact echo signal;

[0131] Determining a time-frequency signal image based on the air-coupled impact echo signal;

[0132] The time-frequency signal image is input into a preset defect detection model, and based on the defect detection model, defect recognition processing is performed on the time-frequency signal image to obtain defect data, wherein the defect detection model is obtained by training the time-frequency signal image samples and the Transformer model block.

[0133] Optionally, before the step of acquiring the air-coupled impact echo signal, the method includes:

[0134] Obtaining a Transformer model block, a time-frequency signal image sample, and a defect data label of the time-frequency signal image sample;

[0135] Based on the Transformer model block, the time-frequency signal image samples and the defect data labels of the time-frequency signal image samples, the model to be trained is trained by a sequence-to-sequence training method to obtain the defect detection model that meets the accuracy conditions.

[0136] Optionally, the step of training the model to be trained by a sequence-to-sequence training method based on the Transformer model block, the time-frequency signal image samples and the defect data labels of the time-frequency signal image samples to obtain the defect detection model that meets the accuracy condition includes:

[0137] Based on the Transformer model block, normalizing the time-frequency signal image samples to obtain normalized time-frequency signal image samples;

[0138] Inputting the normalized time-frequency signal image samples into the model to be trained to obtain predicted defect data;

[0139] Based on the predicted defect data and the defect data labels of the time-frequency signal image samples, the model to be trained is trained by the sequence-to-sequence training method to obtain the defect detection model.

[0140] Optionally, the step of performing defect recognition processing on the time-frequency signal image based on the defect detection model to obtain defect data includes:

[0141] Based on the defect detection model, extracting features from the time-frequency signal image to obtain feature results;

[0142] Based on the characterization results, the defect data is determined.

[0143] Optionally, the step of extracting features from the time-frequency signal image based on the defect detection model to obtain feature results includes:

[0144] Get the target query vector;

[0145] Based on the Transformer model block in the defect detection model, the time-frequency signal image is serialized to obtain each serialization feature;

[0146] Each of the serialized features is fused to obtain a global feature;

[0147] The feature result is obtained based on the global feature and the target query vector.

[0148] Optionally, the step of determining the defect data based on the feature result includes:

[0149] Nonlinear mapping is performed on the feature sequence to obtain the defect data.

[0150] Optionally, the step of determining a time-frequency signal image based on the air-coupled impact echo signal includes:

[0151] Based on the air-coupled impulse echo signal, determining a time domain signal and a frequency domain signal of the air-coupled impulse echo signal;

[0152] The waveform images of the time domain signal and the frequency domain signal are combined to obtain the time-frequency signal image.

[0153] The specific implementation method of the storage medium of the present application is basically the same as the embodiments of the above-mentioned bubble lightweight soil defect detection method, and will not be repeated here.

[0154] The present application also provides a computer program product, including a computer program, which implements the steps of the above-mentioned bubble lightweight soil defect detection method when executed by a processor.

[0155] The specific implementation of the computer program product of the present application is basically the same as the various embodiments of the above-mentioned bubble lightweight soil defect detection method, and will not be repeated here.

[0156] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element.

[0157] The serial numbers of the above-mentioned embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0158] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes a number of instructions for a terminal device (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in each embodiment of the present application.

[0159] The above are only preferred embodiments of the present application, and are not intended to limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the present application specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A method for detecting defects in bubble lightweight soil, characterized in that: The bubble lightweight soil defect detection method comprises: Acquire air-coupled impact echo signal; Determining a time-frequency signal image based on the air-coupled impact echo signal; Input the time-frequency signal image into a preset defect detection model, and perform defect recognition processing on the time-frequency signal image based on the defect detection model to obtain defect data, wherein the defect detection model is obtained by training the time-frequency signal image samples and the Transformer model block; The step of performing defect recognition processing on the time-frequency signal image based on the defect detection model to obtain defect data includes: Based on the defect detection model, extracting features from the time-frequency signal image to obtain feature results; Based on the characteristic result, determining the defect data; The step of extracting features from the time-frequency signal image based on the defect detection model to obtain feature results includes: Get the target query vector; Based on the Transformer model block in the defect detection model, the time-frequency signal image is serialized to obtain each serialization feature; Each of the serialized features is fused to obtain a global feature; The feature result is obtained based on the global feature and the target query vector.

2. The method for detecting defects in bubble lightweight soil according to claim 1, characterized in that: Before the step of acquiring the air-coupled impact echo signal, the method includes: Obtaining a Transformer model block, a time-frequency signal image sample, and a defect data label of the time-frequency signal image sample; Based on the Transformer model block, the time-frequency signal image samples and the defect data labels of the time-frequency signal image samples, the model to be trained is trained by a sequence-to-sequence training method to obtain the defect detection model that meets the accuracy conditions.

3. The method for detecting defects in bubble light soil according to claim 2, characterized in that: The step of training the model to be trained by a sequence-to-sequence training method based on the Transformer model block, the time-frequency signal image samples and the defect data labels of the time-frequency signal image samples to obtain the defect detection model that meets the accuracy condition includes: Based on the Transformer model block, normalizing the time-frequency signal image samples to obtain normalized time-frequency signal image samples; Inputting the normalized time-frequency signal image samples into the model to be trained to obtain predicted defect data; Based on the predicted defect data and the defect data labels of the time-frequency signal image samples, the model to be trained is trained by the sequence-to-sequence training method to obtain the defect detection model.

4. The method for detecting defects in bubble lightweight soil according to claim 1, characterized in that: The step of determining the defect data based on the feature result comprises: The feature sequence is nonlinearly mapped to obtain the defect data.

5. The method for detecting defects in bubble lightweight soil according to claim 1, characterized in that: The step of determining a time-frequency signal image based on the air-coupled impact echo signal comprises: Based on the air-coupled impulse echo signal, determining a time domain signal and a frequency domain signal of the air-coupled impulse echo signal; The waveform images of the time domain signal and the frequency domain signal are combined to obtain the time-frequency signal image.

6. A bubble lightweight soil defect detection device, characterized in that: The bubble lightweight soil defect detection device comprises: An acquisition module, used for acquiring an air-coupled impact echo signal; An image determination module, used to determine a time-frequency signal image based on the air-coupled impact echo signal; an identification module, used for inputting the time-frequency signal image into a preset defect detection model, and performing defect identification processing on the time-frequency signal image based on the defect detection model to obtain defect data, wherein the defect detection model is obtained by training the time-frequency signal image sample and the Transformer model block; The identification module comprises: A feature extraction module, used to extract features from the time-frequency signal image based on the defect detection model to obtain feature results; A defect data determination module, used to determine the defect data based on the feature result; The feature extraction module comprises: A vector acquisition module is used to acquire a target query vector; A serialization module, used for serializing the time-frequency signal image based on the Transformer model block in the defect detection model to obtain each serialization feature; A feature fusion module, used for fusing each of the serialized features to obtain a global feature; A result determination module is used to obtain the feature result based on the global feature and the target query vector.

7. A bubble lightweight soil defect detection device, characterized in that: The bubble light soil defect detection device comprises: a memory, a processor, and a program stored in the memory for implementing the bubble light soil defect detection method. The memory is used to store a program for implementing a method for detecting defects in air bubble lightweight soil; The processor is used to execute a program for implementing the method for detecting defects in bubble lightweight soil, so as to implement the steps of the method for detecting defects in bubble lightweight soil as described in any one of claims 1 to 5.

8. A storage medium, characterized in that: The storage medium stores a program for implementing a method for detecting defects in bubble lightweight soil, and the program for implementing a method for detecting defects in bubble lightweight soil is executed by a processor to implement the steps of the method for detecting defects in bubble lightweight soil as described in any one of claims 1 to 5.

Images