Method and device for detecting loosening state of threaded connection structure based on knocking method
By generating time-frequency graphs through variational mode decomposition, compression reconstruction, and short-time Fourier transform, and combining them with an improved convolutional neural network, the accuracy and applicability issues of loose threaded connection structure detection in existing technologies are solved, achieving efficient automatic detection.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- BEIJING INST OF TECH TANGSHAN RES INST
- Filing Date
- 2023-10-07
- Publication Date
- 2026-07-21
AI Technical Summary
In the existing technology, sensor and vision-based methods for detecting loose threaded connections are either costly or have low accuracy, while tapping-based methods rely on human experience, resulting in low detection accuracy and making them unsuitable for early loosening detection.
Variational mode decomposition and compression reconstruction techniques are used to process the signal in the frequency and time domains. A time-frequency graph is generated by combining short-time Fourier transform, and an improved convolutional neural network is used for feature extraction and classification to achieve automatic detection of loose threaded connection structures.
It enables the detection of threaded connection structures in different positions and under different loosening conditions, improving the accuracy and applicability of the detection and reducing the reliance on human experience.
Smart Images
Figure CN117388383B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of health monitoring technology for threaded connection structures, and in particular to a method and apparatus for detecting the loosening state of threaded connection structures based on the tapping method. Background Technology
[0002] In engineering, threaded connections are widely used in industrial fields due to their advantages such as ease of assembly and disassembly, reusability, and simple operation. However, under the influence of various external factors, threaded connections inevitably experience phenomena such as a decrease in preload and loosening. Therefore, it is necessary to periodically inspect threaded connections for loosening.
[0003] There are three main methods for detecting looseness in traditional threaded connections: sensor-based, vision-based, and tapping-based methods. However, sensor-based methods are not only costly and complex, making them difficult to implement on a large scale. Vision-based methods are not sensitive to minor loosening and are not suitable for early-stage loosening. The tapping method relies on experienced workers to distinguish the tapping sounds to determine if the threaded connection is loose, resulting in low accuracy and a high dependence on worker experience. Summary of the Invention
[0004] To address the aforementioned problems, embodiments of the present invention provide a method and apparatus for detecting the loosening state of threaded connection structures based on the tapping method.
[0005] One aspect of this invention provides a method for detecting the loosening state of a threaded connection structure based on a tapping method, comprising:
[0006] Collect the sound signal generated when the threaded connection structure is struck;
[0007] The first algorithm is used to decompose the collected sound signal to obtain multiple intrinsic mode functions, each of which represents a signal component with a center frequency.
[0008] The target intrinsic mode function is selected based on the center frequency of each intrinsic mode function, and the target intrinsic mode function is reconstructed to obtain the denoised signal;
[0009] The sparse signal is obtained by using the sparse representation of the denoised signal in the second algorithm, and the sparse signal is fused with the denoised signal to generate a fused signal.
[0010] The fused signal is converted into a time-frequency graph using a short-time Fourier transform. The time-frequency graph is then input into a pre-trained convolutional neural network for loosening state detection to obtain the loosening state of the threaded connection structure.
[0011] Optionally, the process of training the loose state detection convolutional neural network includes:
[0012] The sound signals generated when the threaded connection structure is struck at different locations and in different loose states are collected as sample data.
[0013] The sample data is preprocessed to obtain training samples. An initial convolutional neural network is built and trained using the training samples to obtain a loose state detection convolutional neural network.
[0014] Optionally, the initial convolutional neural network is constructed by inserting a CBAM attention mechanism network into the traditional DenseNet-121 network: Specifically, the sample data first passes through a 7×7 convolutional layer and a 3×3 max pooling layer; then the data enters the first Dense block, which contains 6 bottlenecks, each consisting of a 1×1 convolutional layer and a 3×3 convolutional layer; then the data enters the first Transition layer, each containing a 1×1 convolutional layer and a 2×2 average pooling layer; subsequently, it passes through the second Dense block, the first CBAM attention mechanism network, the second Transition layer, the third Dense block, the third Transition layer, the fourth Dense block, and the second CBAM attention mechanism network in sequence, before passing through a 7×7 average pooling layer and a fully connected layer to obtain the output result.
[0015] Optionally, the bottleneck structures in each Dense block are the same but the number varies; the structures in each Transition layer are the same; the CBAM attention mechanism network takes the output of the Dense block as input, and performs max pooling and average pooling operations on each channel respectively. The two channel features obtained after pooling are fed into a fully connected layer for learning, and finally the two channel features are added to generate a one-dimensional channel attention; the generated one-dimensional channel attention is multiplied with the input to generate input features; the input features are subjected to max pooling and average pooling operations on the spatial aspect, and the two two-dimensional vectors generated by pooling are concatenated and then convolved to generate a two-dimensional spatial attention; the two-dimensional spatial attention and the channel attention are multiplied element-wise to obtain the final output result.
[0016] Optionally, the acquired sound signal can be decomposed according to different center frequencies using a variational mode decomposition algorithm to obtain multiple intrinsic mode functions.
[0017] Optionally, the process of selecting a target intrinsic mode function based on the center frequency of each intrinsic mode function and reconstructing the target intrinsic mode function includes:
[0018] Observe the frequency domain of the collected sound signal and determine the concentration range of characteristic frequencies in the sound signal based on the frequency domain.
[0019] Based on the determined range of characteristic frequencies, the intrinsic mode function corresponding to the center frequency within the range is selected as the target intrinsic mode function, and the selected target intrinsic mode functions are superimposed.
[0020] Optionally, the denoised signal is represented by a sparse representation based on compression reconstruction technology. In the compression reconstruction technology, the sparse basis is selected as the discrete Fourier transform matrix, the measurement matrix is selected as the random Gaussian matrix, the sparse matrix and the measurement matrix are multiplied to form the sensing matrix, and the signal reconstruction algorithm is selected as the compressed sampling matching pursuit algorithm.
[0021] Another aspect of this invention provides a device for detecting the loosening state of a threaded connection structure based on a tapping method, comprising:
[0022] The signal preprocessing module is used to collect the sound signal generated when the threaded connection structure is struck; and to decompose the collected sound signal using the first algorithm to obtain multiple intrinsic mode functions. Each intrinsic mode function represents a signal component of a different frequency. The target intrinsic mode function is selected according to the center frequency of each intrinsic mode function and the target intrinsic mode function is reconstructed to obtain a denoised signal.
[0023] The loosening state information enhancement module uses the second algorithm to denoise the signal to obtain a sparse signal, and then fuses the sparse signal with the denoised signal to generate a fused signal. The fused signal is then converted into a time-frequency graph using a short-time Fourier transform.
[0024] The loosening state classification module inputs the time-frequency graph into a pre-trained loosening state detection convolutional neural network to obtain the loosening state of the threaded connection structure.
[0025] In another aspect of the present invention, an electronic device is provided, characterized in that it includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus.
[0026] Memory is used to store processor-executable instructions;
[0027] The processor, when executing instructions stored in memory, implements the aforementioned method for detecting the loosening state of threaded connection structures based on the tapping method.
[0028] Compared with the prior art, the beneficial effects of the present invention are as follows: The present invention uses variational mode decomposition to denoise the signal in the frequency domain, and uses compression reconstruction technology to enhance the signal in the time domain. The processed signal is then transformed by short-time Fourier transform to generate a time-frequency diagram. Finally, an improved convolutional neural network architecture is used to extract features and perform classification learning to detect the loose state of threaded connection structures. This invention is applicable to the detection of threaded connection structures in different positions and under different loose states. Attached Figure Description
[0029] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, are not intended to limit the scope of the invention. In the drawings:
[0030] Figure 1 A flowchart illustrating a method for detecting the loosening state of a threaded connection structure based on the tapping method, provided in an embodiment of the present invention;
[0031] Figure 2 A schematic diagram of a loosening detection device for a threaded connection structure based on the tapping method provided in an embodiment of the present invention;
[0032] Figure 3 This is a schematic diagram of an electronic device structure provided in an embodiment of the present invention. Detailed Implementation
[0033] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the embodiments and accompanying drawings. Here, the illustrative embodiments and descriptions of this invention are used to explain the invention, but are not intended to limit the invention.
[0034] See Figure 1 The present invention provides a method for detecting the loosening state of a threaded connection structure based on the tapping method, comprising:
[0035] S1. Collect the sound signal generated when the threaded connection structure is struck.
[0036] In practice, a hammer can be used to strike the threaded connection structure, while a suitable sensor or microphone is used to collect the sound generated during the striking process. The sound is then used as an input signal, and a dual-threshold detection method based on short-time energy and zero-crossing rate is employed to perform endpoint detection and separate the desired sound signal. Using a dual-threshold detection method based on short-time energy and zero-crossing rate avoids the confusion between unvoiced and voiced segments when using only the short-time energy threshold, and the confusion between audible and silent segments when using only the short-time zero-crossing rate threshold.
[0037] The dual-threshold detection method based on short-time energy and zero-crossing rate requires setting three thresholds for the audio signal: a high threshold for short-time energy, a low threshold for short-time energy, and a threshold for short-time zero-crossing rate. The input signal is then segmented into frames, and the energy and zero-crossing rate of each frame are calculated. The starting point of the input signal is then determined: if the energy of a frame exceeds the low threshold for short-time energy or the zero-crossing rate exceeds the short-time zero-crossing rate threshold, it is considered the start of the input signal. When the energy of subsequent frames exceeds the high threshold for short-time energy, the audio phase is considered to have officially begun. If this phase is maintained for a certain duration, this segment of signal is considered the desired audio signal.
[0038] S2. The first algorithm is used to decompose the collected sound signal to obtain multiple intrinsic mode functions, each of which represents a signal component with a center frequency.
[0039] In practice, variational mode decomposition algorithms can be used to decompose the acquired sound signal according to different center frequencies to obtain multiple intrinsic mode functions.
[0040] Variational mode decomposition (VMD) algorithms decompose sound signals into multiple discrete sub-signals, each of which is a modal component signal revolving around its own center frequency. The VMD algorithm determines the number of modal decompositions of the signal based on the actual situation, and adaptively matches the center frequency and finite bandwidth of each mode in the subsequent iterative solution process, finally completing the decomposition of the signal to obtain multiple intrinsic mode functions.
[0041] S3. Select the target intrinsic mode function based on the center frequency of each intrinsic mode function and reconstruct the target intrinsic mode function to obtain the denoised signal.
[0042] In practice, the frequency domain of the collected sound signal can be observed, and the concentration range of characteristic frequencies in the sound signal can be determined based on the frequency domain. Based on the determined concentration range of characteristic frequencies, the intrinsic mode function corresponding to the center frequency within the concentration range is selected as the target intrinsic mode function, and the selected target intrinsic mode functions are superimposed to obtain the denoised signal.
[0043] S4. Use the second algorithm to sparsely represent the denoised signal to obtain a sparse signal, and then fuse the sparse signal with the denoised signal to generate a fused signal.
[0044] In implementation, the denoised signal can be represented sparsely using compression reconstruction technology. In this technology, the sparse basis is chosen as the Discrete Fourier Transform matrix, the measurement matrix as the random Gaussian matrix, and the sparse matrix and measurement matrix are multiplied to form the sensing matrix. The signal reconstruction algorithm chosen is the compressed sampling matched pursuit algorithm. Furthermore, the addWeighted function is selected for the fusion operation.
[0045] S5. Use short-time Fourier transform to convert the fused signal into a time-frequency graph, and input the time-frequency graph into a pre-trained loose state detection convolutional neural network to obtain the loose state of the threaded connection structure.
[0046] In practice, the process of training the loose state detection convolutional neural network includes:
[0047] The sound signals generated when tapping threaded connection structures located in different parts and in different loose states are collected as sample data. For example, the sound signals generated when tapping the bolts of flange connections, the threads of pipe joints, and the bolts in lap joint structures can be collected to increase the richness of the sample and make it applicable to the detection of the loose state of threaded connection structures under various working conditions.
[0048] The sample data is preprocessed to obtain training samples. The preprocessing process is the same as the above-mentioned noise reduction and enhancement process of the sound signal based on variational mode decomposition algorithm and compression reconstruction technology. An initial convolutional neural network is built and trained using the training samples to obtain a loose state detection convolutional neural network.
[0049] In practice, the initial convolutional neural network is built by inserting a CBAM attention mechanism network on top of the traditional DenseNet-121 network. Specifically, the sample data first passes through a 7×7 convolutional layer and a 3×3 max pooling layer. Then, the data enters the first Dense block, which contains 6 bottlenecks, each consisting of a 1×1 convolutional layer and a 3×3 convolutional layer. After that, the data enters the first Transition layer, each containing a 1×1 convolutional layer and a 2×2 average pooling layer. Subsequently, the data passes through the second Dense block, the first CBAM attention mechanism network, the second Transition layer, the third Dense block, the third Transition layer, the fourth Dense block, and the second CBAM attention mechanism network in sequence, before passing through a 7×7 average pooling layer and a fully connected layer to obtain the output result.
[0050] In this network, each Dense block has the same bottleneck structure but a different number of bottlenecks; each Transition layer has the same structure; the CBAM attention mechanism takes the output of the Dense block as input, and performs max pooling and average pooling operations on each channel respectively. The two channel features obtained after pooling are fed into a fully connected layer for learning, and finally the two channel features are added to generate a one-dimensional channel attention; the generated one-dimensional channel attention is multiplied with the input to generate the input features; the input features are subjected to max pooling and average pooling operations on the spatial dimension, and the two two-dimensional vectors generated by pooling are concatenated and then convolved to generate a two-dimensional spatial attention; the two-dimensional spatial attention and the channel attention are multiplied element-wise to obtain the final output.
[0051] The solution provided by this invention uses variational mode decomposition to denoise the signal in the frequency domain, while compression reconstruction technology is used to enhance the signal in the time domain. Short-time Fourier transform is then used to perform time-frequency transformation on the processed signal to generate a time-frequency graph. Finally, an improved convolutional neural network architecture is used to extract features and perform classification learning to detect the loose state of threaded connection structures. This method is applicable to the detection of threaded connection structures in different positions and under different loose states.
[0052] See Figure 2 In another aspect of this invention, a device for detecting the loosening state of a threaded connection structure based on a tapping method is provided, comprising:
[0053] The signal preprocessing module 200 is used to collect the sound signal generated when the threaded connection structure is struck; and to decompose the collected sound signal using the first algorithm to obtain multiple intrinsic mode functions. Each intrinsic mode function represents a signal component of a different frequency. The target intrinsic mode function is selected according to the center frequency of each intrinsic mode function. The selected target intrinsic mode functions are superimposed to obtain a denoised signal.
[0054] The loosening state information enhancement module 210 uses the second algorithm to denoise the signal to obtain a sparse signal, and then fuses the sparse signal with the denoised signal to generate a fused signal. The fused signal is then converted into a time-frequency diagram using a short-time Fourier transform.
[0055] The loosening state classification module 220 inputs the time-frequency map into the pre-trained loosening state detection convolutional neural network to obtain the loosening state of the threaded connection structure.
[0056] The solution provided by this invention uses variational mode decomposition to denoise the signal in the frequency domain, while compression reconstruction technology is used to enhance the signal in the time domain. Short-time Fourier transform is then used to perform time-frequency transformation on the processed signal to generate a time-frequency graph. Finally, an improved convolutional neural network architecture is used to extract features and perform classification learning to detect the loose state of threaded connection structures. This method is applicable to the detection of threaded connection structures in different positions and under different loose states.
[0057] This invention also provides an electronic device, such as... Figure 3 As shown, it includes a processor 001, a communication interface 002, a memory 003, and a communication bus 004. The processor 001, communication interface 002, and memory 003 communicate with each other via the communication bus 004.
[0058] Memory 003 is used to store computer programs;
[0059] Processor 001, when executing the program stored in memory 003, implements the above-mentioned method for detecting the loosening state of threaded connection structures based on the tapping method, including:
[0060] Collect the sound signal generated when the threaded connection structure is struck;
[0061] The first algorithm is used to decompose the collected sound signal to obtain multiple intrinsic mode functions, each of which represents a signal component with a center frequency.
[0062] The target intrinsic mode function is selected based on the center frequency of each intrinsic mode function, and the target intrinsic mode function is reconstructed to obtain the denoised signal;
[0063] The sparse signal is obtained by using the sparse representation of the denoised signal in the second algorithm, and the sparse signal is fused with the denoised signal to generate a fused signal.
[0064] The fused signal is converted into a time-frequency graph using a short-time Fourier transform. The time-frequency graph is then input into a pre-trained convolutional neural network for loosening state detection to obtain the loosening state of the threaded connection structure.
[0065] The solution provided by this invention uses variational mode decomposition to denoise the signal in the frequency domain, while compression reconstruction technology is used to enhance the signal in the time domain. Short-time Fourier transform is then used to perform time-frequency transformation on the processed signal to generate a time-frequency graph. Finally, an improved convolutional neural network architecture is used to extract features and perform classification learning to detect the loose state of threaded connection structures. This method is applicable to the detection of threaded connection structures in different positions and under different loose states.
[0066] The communication bus mentioned in the above electronic devices can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. This communication bus can be divided into address bus, data bus, control bus, etc. For ease of illustration, only one thick line is used to represent it in the diagram, but this does not mean that there is only one bus or one type of bus.
[0067] The communication interface is used for communication between the aforementioned electronic devices and other devices.
[0068] The memory may include random access memory (RAM) or non-volatile memory (NVM), such as at least one disk storage device. Optionally, the memory may also be at least one storage device located remotely from the aforementioned processor.
[0069] The processors mentioned above can be general-purpose processors, including central processing units (CPUs), network processors (NPs), etc.; they can also be digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.
[0070] In the above embodiments, implementation can be achieved entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid state disk (SSD)).
[0071] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, 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 such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.
[0072] The various embodiments in this specification are described in a related manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the device and electronic device embodiments are basically similar to the method embodiments, so the descriptions are relatively simple; relevant parts can be referred to the descriptions of the method embodiments.
[0073] The above description is merely a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention are included within the scope of protection of the present invention.
Claims
1. A method for detecting the loosening state of a threaded connection structure based on the tapping method, characterized in that, include: Collect the sound signal generated when the threaded connection structure is struck; The variational mode decomposition algorithm is used to decompose the acquired sound signal according to different center frequencies to obtain multiple intrinsic mode functions. Each intrinsic mode function represents a signal component with a center frequency. The target intrinsic mode function is selected based on the center frequency of each intrinsic mode function, and the target intrinsic mode function is reconstructed to obtain the denoised signal; The sparse signal is obtained by sparse characterization of the denoised signal based on compression reconstruction technology, and the sparse signal is fused with the denoised signal to generate a fused signal. In the compression reconstruction technology, the sparse basis is selected as the discrete Fourier transform matrix, the measurement matrix is selected as the random Gaussian matrix, the sparse matrix and the measurement matrix are multiplied to form the sensing matrix, and the signal reconstruction algorithm is selected as the compressed sampling matching pursuit algorithm. The fused signal is converted into a time-frequency graph using a short-time Fourier transform. The time-frequency graph is then input into a pre-trained convolutional neural network for loosening state detection to obtain the loosening state of the threaded connection structure.
2. The method for detecting looseness in threaded connections based on the tapping method as described in claim 1, characterized in that, The process of training a loose state detection convolutional neural network includes: The sound signals generated when the threaded connection structure is struck at different locations and in different loose states are collected as sample data. The sample data is preprocessed to obtain training samples. An initial convolutional neural network is built and trained using the training samples to obtain a loose state detection convolutional neural network.
3. The method for detecting looseness in threaded connections based on the tapping method as described in claim 2, characterized in that, The initial convolutional neural network was built by inserting a CBAM attention mechanism network into the traditional DenseNet-121 network: Specifically, the sample data first passes through a 7×7 convolutional layer and a 3×3 max pooling layer; then the data enters the first Dense block, which contains 6 bottlenecks, each consisting of a 1×1 convolutional layer and a 3×3 convolutional layer; then the data enters the first Transition layer, each containing a 1×1 convolutional layer and a 2×2 average pooling layer; subsequently, it passes through the second Dense block, the first CBAM attention mechanism network, the second Transition layer, the third Dense block, the third Transition layer, the fourth Dense block, and the second CBAM attention mechanism network in sequence, before passing through a 7×7 average pooling layer and a fully connected layer to obtain the output result.
4. The method for detecting looseness in threaded connections based on the tapping method as described in claim 3, characterized in that, Each Dense block has the same bottleneck structure but a different number of bottlenecks; each Transition layer has the same structure. The CBAM attention mechanism network takes the output of the Dense block as input and performs max pooling and average pooling operations on each channel. The two channel features obtained after pooling are fed into a fully connected layer for learning. Finally, the two channel features are added to generate a one-dimensional channel attention. The generated one-dimensional channel attention is multiplied with the input to generate the input features. The input features are then subjected to max pooling and average pooling operations on the spatial dimension. The two two-dimensional vectors generated by pooling are concatenated and then convolved to generate a two-dimensional spatial attention. The two-dimensional spatial attention and the channel attention are multiplied element-wise to obtain the final output.
5. The method for detecting looseness in threaded connections based on the tapping method as described in claim 1, characterized in that, The process of selecting the target intrinsic mode function based on the center frequency of each intrinsic mode function and reconstructing the target intrinsic mode function includes: Observe the frequency domain of the collected sound signal and determine the concentration range of characteristic frequencies in the sound signal based on the frequency domain. Based on the determined range of characteristic frequencies, the intrinsic mode function corresponding to the center frequency within the range is selected as the target intrinsic mode function, and the selected target intrinsic mode functions are superimposed.
6. A device for detecting the loosening state of a threaded connection structure based on a tapping method, characterized in that, include: The signal preprocessing module is used to collect the sound signals generated when the threaded connection structure is struck. The acquired sound signal is decomposed according to different center frequencies using the variational mode decomposition algorithm to obtain multiple intrinsic mode functions. Each intrinsic mode function represents a signal component of different frequencies. The target intrinsic mode function is selected according to the center frequency of each intrinsic mode function and reconstructed to obtain the denoised signal. The loosening state information enhancement module uses compression reconstruction technology to sparsely represent the denoised signal to obtain a sparse signal, and then fuses the sparse signal with the denoised signal to generate a fused signal. The fused signal is converted into a time-frequency diagram using short-time Fourier transform. In the compression reconstruction technology, the sparse basis is selected as the discrete Fourier transform matrix, the measurement matrix is selected as the random Gaussian matrix, and the sparse matrix and the measurement matrix are multiplied to form the sensing matrix. The signal reconstruction algorithm is the compressed sampling matching pursuit algorithm. The loosening state classification module inputs the time-frequency graph into a pre-trained loosening state detection convolutional neural network to obtain the loosening state of the threaded connection structure.
7. An electronic device, characterized in that, It includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; Memory is used to store processor-executable instructions; A processor, when executing instructions stored in memory, implements the steps of the method described in any one of claims 1-5.