Ultrasonic time-frequency joint analysis method used under non-coupling condition

Abstract

The invention provides an ultrasonic time-frequency joint analysis method used under a non-coupling condition. The method comprises the following steps: transmitting 1MHz ultrasonic waves and receiving reflected signals through an air coupling technology; after the signals are preprocessed, the signal-to-noise ratio is improved by 18.81 dB through wavelet filtering, and then features are extracted through time-frequency analysis. And the multi-scale convolutional neural network performs semantic segmentation on the features, outputs a high-resolution defect distribution map, and enhances the defect identification precision in combination with a time reversal detection technology. And finally, fusing the segmentation result and the inversion signal to realize quantitative analysis of the hollowing and the crack of the wall body. According to the method, the defect detection precision can be improved, efficient nondestructive detection without couplant residues is realized, the phi 5mm-level defects are accurately identified, and the incidence relation between the hollowing volume and the crack depth is quantified.

Description

Technical Field

[0001] The present invention relates to the field of non-destructive testing technology, and more particularly to an ultrasonic time-frequency joint analysis method for use under non-coupling conditions. Background Art

[0002] In the field of ultrasonic testing, coupling agent, as an important medium between the probe and the object being tested, has always been a key component of traditional ultrasonic flaw detection technology. The main function of the coupling agent is to fill the tiny gap between the probe and the surface of the object being tested, exclude air, and reduce the reflection loss of sound waves at the interface, thereby ensuring that the sound waves can be effectively transmitted from the probe to the material being tested. However, the use of coupling agent also brings many problems. For example, the residual coupling agent on the surface of the object being tested can negatively affect the subsequent coating system. Especially in high-salt environments such as near-shore ports, the residual coupling agent can absorb moisture, creating a micro-corrosive environment, resulting in reduced coating adhesion and premature failure of anti-corrosion properties. In addition, the use of coupling agent increases the complexity and cost of testing, requiring cleaning and maintenance before and after testing.

[0003] In recent years, with the development of technology, air-coupled ultrasonic testing technology has gradually attracted attention. This technology directly uses air as the coupling medium, eliminating the need for traditional coupling agents and thus avoiding the problems caused by coupling agent residue. However, air-coupled ultrasonic testing technology also faces certain challenges. Due to the significant difference in acoustic impedance between air and solid materials, ultrasonic waves suffer significant reflection losses at the air-solid interface, resulting in weak signal strength and a low signal-to-noise ratio. In addition, traditional signal processing methods have difficulty effectively extracting significant features from air-coupled ultrasonic signals, limiting the application of this technology in practical engineering projects.

[0004] In the process of implementing the embodiments of the present invention, there are at least the following problems or defects in the existing technology: on the one hand, traditional ultrasonic detection technology relies on coupling agents, which not only increases the complexity and cost of detection, but may also cause corrosion and failure of subsequent coating systems; on the other hand, although the existing air-coupled ultrasonic detection technology avoids the use of coupling agents, it has deficiencies in signal strength and signal-to-noise ratio, and lacks effective signal processing methods to extract defect characteristics, making it difficult to meet the needs of high-precision detection. Summary of the Invention

[0005] The present invention provides a method for ultrasonic time-frequency joint analysis under uncoupled conditions, comprising: Step 1: Use an air-coupled ultrasonic detection device to transmit an ultrasonic pulse signal with a frequency of 1 MHz to the wall under test. The air-coupled ultrasonic detection device uses air as a coupling medium, and the transmitted ultrasonic wave penetrates the wall under test; Step 2: Receive the reflected ultrasonic signal through the receiving probe of the air-coupled ultrasonic detection device to obtain the original received signal; Step 3: Inputting the original received signal into the receiving driving circuit for signal preprocessing, wherein the signal preprocessing includes signal amplification, signal shaping and detection filtering, and outputting the preprocessed signal; Step 4: Perform wavelet filtering on the preprocessed signal to obtain a denoised signal, which includes: Sub-step 4.1: Perform wavelet transform on the preprocessed signal to obtain a wavelet domain signal; Sub-step 4.2: Separate the effective signal coefficient and noise coefficient in the wavelet domain signal; Sub-step 4.3: discard the noise coefficient; Sub-step 4.4: Perform inverse wavelet transform on the effective signal coefficients to reconstruct the denoised signal; Step 5: Perform a time-frequency joint analysis operation on the denoised signal to extract the time-frequency domain feature signal; Step 6: Input the time-frequency domain feature signal into a multi-scale convolutional neural network to perform a semantic image segmentation operation, and output a defect area segmentation map, wherein the input image and the output segmentation map of the multi-scale convolutional neural network have the same resolution; Step 7: Perform time reversal detection processing on the time-frequency domain feature signal to obtain a time reversal signal; Step 8: Generate quantitative detection results of wall defects based on the defect area segmentation map and time reversal signal.

[0006] Furthermore, the wavelet filtering process in step 4 satisfies the signal-to-noise ratio improvement relationship: in is the signal-to-noise ratio of the preprocessed signal, is the signal-to-noise ratio of the denoised signal.

[0007] Furthermore, the semantic image segmentation operation in step 6 includes the following sub-steps: Sub-step 6.1: Perform a padding operation on the input time-frequency domain feature signal to obtain a padded feature map; Sub-step 6.2: Perform convolution operations on the padded feature map to extract vertical and horizontal features; Sub-step 6.3: Perform a maximum pooling operation on the feature map output by the convolution operation to obtain compressed feature data; Sub-step 6.4: Output the defect area segmentation map.

[0008] Furthermore, the method further includes step nine: Perform spatiotemporal alignment on the RGB image V and the infrared thermal image T, and optimize the modal weight coefficients through non-negative matrix decomposition , establish a multimodal fusion model: in, represents the tensor product, is the fused feature output, is an RGB image, For infrared thermal imaging, is the modal weight coefficient, is the number of modes, The first modal data, For infrared thermal imaging modal data.

[0009] Furthermore, in step 1, the propagation of ultrasonic waves satisfies the wave equation: in, is the density of the wall material being measured, is the particle displacement, is the elastic coefficient, For time, is the spatial coordinate.

[0010] Furthermore, the time reversal detection process in step 7 satisfies: in, is the input time-frequency domain characteristic signal, is the time-reversed signal output, For time, is the attenuation compensation factor under air coupling conditions.

[0011] Furthermore, the driving circuit of the air-coupled ultrasonic detection device includes: Oscillation source module, used to generate high-frequency square wave signal; The frequency division circuit module uses a dual D flip-flop CD4013 to perform frequency division processing on the high-frequency square wave signal; The power amplifier module uses the IRF830 amplifier to amplify the power of the divided signal; The non-polarized capacitive coupling module couples the power-amplified signal to the transmitting probe.

[0012] Furthermore, the defect quantitative analysis operation in step eight includes: Sub-step 8.1: performing a C-scan image reconstruction operation on the defect region segmentation map to output a defect spatial distribution map; Sub-step 8.2: Calculating the transmitted wave group velocity based on the time-reversed signal; Sub-step 8.3: Invert the exponential relationship between the surface crack depth d and the internal hollow volume V based on the transmitted wave group velocity: , where V is the internal hollow volume; k is the material constant, which is a fixed coefficient related to the wall material properties; c is the material constant, which reflects the rate at which the internal hollow volume changes with the depth of the surface crack; d is the depth of the surface crack; Sub-step 8.4: Mark the defect area with a diameter of 5 mm in the defect spatial distribution map.

[0013] Furthermore, in step 1, the propagation of ultrasonic waves in the solid medium satisfies the isotropic wave equation: in, is the density of the wall material being measured, u is the particle displacement, t is the time, x and y are the spatial coordinates, and is the Mellar constant.

[0014] Furthermore, the signal preprocessing in step 3 also includes a low-pass filtering operation. When there is no external force, the sound field equation is: And the signal enhancement processing satisfies wavelet reconstruction: in, is the density of the wall material being measured, u is the displacement of the mass point, t is the time, C is the elastic coefficient, is the Laplace operator, For the enhanced signal output, is the wavelet transform coefficient, a is the scale parameter, b is the translation parameter, is the wavelet basis function.

[0015] The above embodiments of the present invention have at least the following beneficial effects: 1. By adopting air-coupled ultrasonic detection technology, the problem of reduced coating anti-corrosion performance caused by traditional coupling agent residue is effectively solved, the corrosion effect of coupling agent residue on the substrate surface is avoided, and the risk of reduced coating adhesion caused by incomplete coupling agent cleaning is eliminated. It is particularly suitable for building structure inspection scenarios that require long-term anti-corrosion protection.

[0016] 2. The innovative signal processing solution combines wavelet filtering and time-frequency joint analysis methods to significantly improve the sensitivity and accuracy of defect detection. It can effectively identify tiny defect characteristics that are difficult to detect with traditional methods. It solves the technical difficulties of severe signal attenuation and low signal-to-noise ratio in conventional ultrasonic detection in complex building structures, and provides more reliable detection data for building structure health assessment.

[0017] 3. The intelligent analysis system based on multi-scale convolutional neural networks and time reversal technology realizes the automatic identification and quantitative evaluation of defects, overcoming the subjectivity and instability of manual interpretation. At the same time, by establishing a correlation model between hollow volume and crack depth, it provides a scientific basis for damage assessment and maintenance decision-making of building structures, and solves the pain point of insufficient quantitative analysis of traditional detection methods. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily apparent by reading the following detailed description with reference to the accompanying drawings, in which several embodiments of the present invention are shown by way of example and not limitation, in which: Figure 1 A flowchart of a method for joint time-frequency analysis of ultrasound under uncoupled conditions provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0019] The principles and spirit of the present invention will be described below with reference to several exemplary embodiments. It should be understood that these embodiments are provided solely to enable those skilled in the art to better understand and implement the present invention, and are not intended to limit the scope of the present invention in any way. Rather, these embodiments are provided to make the present invention more thorough and complete, and to fully convey the scope of the present invention to those skilled in the art.

[0020] Those skilled in the art will appreciate that the embodiments of the present invention may be implemented as a system, apparatus, device, method, or computer program product. Therefore, the present invention may be implemented in the following forms: entirely in hardware, entirely in software (including firmware, resident software, microcode, etc.), or in a combination of hardware and software.

[0021] It should be noted that any number of elements in the drawings is for illustration only and not for limitation, and any naming is only for distinction and does not have any limiting meaning.

[0022] Reference below Figure 1 , Figure 1 This is a flow chart of a method for ultrasonic time-frequency joint analysis under uncoupled conditions provided by one embodiment of the present invention. Figure 1 As shown, a method for ultrasonic time-frequency joint analysis under uncoupled conditions includes: Step 1: Use an air-coupled ultrasonic detection device to transmit an ultrasonic pulse signal with a frequency of 1 MHz to the wall under test. The air-coupled ultrasonic detection device uses air as a coupling medium, and the transmitted ultrasonic wave penetrates the wall under test; Step 2: Receive the reflected ultrasonic signal through the receiving probe of the air-coupled ultrasonic detection device to obtain the original received signal; Step 3: Inputting the original received signal into the receiving driving circuit for signal preprocessing, wherein the signal preprocessing includes signal amplification, signal shaping and detection filtering, and outputting the preprocessed signal; Step 4: Perform wavelet filtering on the preprocessed signal to obtain a denoised signal, which includes: Sub-step 4.1: Perform wavelet transform on the preprocessed signal to obtain a wavelet domain signal; Sub-step 4.2: Separate the effective signal coefficient and noise coefficient in the wavelet domain signal; Sub-step 4.3: discard the noise coefficient; Sub-step 4.4: Perform inverse wavelet transform on the effective signal coefficients to reconstruct the denoised signal; Step 5: Perform a time-frequency joint analysis operation on the denoised signal to extract the time-frequency domain feature signal; Step 6: Input the time-frequency domain feature signal into a multi-scale convolutional neural network to perform a semantic image segmentation operation, and output a defect area segmentation map, wherein the input image and the output segmentation map of the multi-scale convolutional neural network have the same resolution; Step 7: Perform time reversal detection processing on the time-frequency domain feature signal to obtain a time reversal signal; Step 8: Generate quantitative detection results of wall defects based on the defect area segmentation map and time reversal signal.

[0023] It should be noted that the present invention proposes an ultrasonic time-frequency joint analysis method for uncoupled conditions. This method transmits ultrasonic pulse signals through an air-coupled ultrasonic detection device and receives reflected signals. After a series of signal processing steps, it finally realizes the quantitative detection of wall defects. The air-coupled ultrasonic detection device is a detection device that uses air as a coupling medium. It does not require the use of traditional coupling agents, thereby avoiding the influence of coupling agent residues on subsequent coating systems. In the signal processing process, wavelet filtering is a filtering method based on wavelet transform. It significantly improves the signal-to-noise ratio by decomposing the signal into different frequency components and removing the noise coefficient. The time-frequency joint analysis operation further enhances the effective feature extraction rate of the signal by extracting feature signals in the time-frequency domain. The multi-scale convolutional neural network CNN is a deep learning model used to perform semantic image segmentation on feature signals in the time-frequency domain and output a defect area segmentation map with the same resolution as the input image, thereby realizing accurate positioning and identification of wall defects.

[0024] Specifically, the ultrasonic pulse signal emitted by the air-coupled ultrasonic detection device has a frequency of 1 MHz. This high-frequency signal can penetrate walls and detect reflected signals. The reflected signal received by the receiving probe undergoes preprocessing by the receiving driver circuit, including signal amplification, shaping, and detection filtering. These operations enhance signal strength and remove some noise. The wavelet transform operation in wavelet filtering converts the signal from the time domain to the wavelet domain, separates the effective signal coefficient from the noise coefficient, discards the noise coefficient, and then reconstructs the denoised signal through an inverse wavelet transform. The joint time-frequency analysis operation extracts characteristic signals that characterize wall defects by analyzing the signal's variations at different times and frequencies. The multi-scale convolutional neural network takes the time-frequency domain feature signals as input and outputs a segmentation map of the defect region. The network extracts vertical and horizontal features through convolution operations and compresses the feature data through maximum pooling. The final output is a segmentation map with the same resolution as the input image, where each pixel is labeled with its corresponding defect category.

[0025] Preferably, the driving circuit of the air-coupled ultrasonic detection device includes an oscillation source module, a frequency division circuit module, a power amplifier module and a non-polar capacitor coupling module. The oscillation source module is used to generate a high-frequency square wave signal, the frequency division circuit module uses a dual D flip-flop CD4013 to perform frequency division processing on the high-frequency square wave signal, the power amplifier module uses an IRF830 amplifier to power amplify the frequency-divided signal, and the non-polar capacitor coupling module couples the power-amplified signal to the transmitting probe. In the signal preprocessing step, the low-pass filtering operation is used to further remove high-frequency noise and ensure the purity of the signal. In the defect quantitative analysis operation, the defect spatial distribution map is output through the C-scan image reconstruction operation, the transmitted wave group velocity is calculated based on the time-reversed signal, and the exponential relationship between the surface crack depth and the internal hollow volume is inverted based on the transmitted wave group velocity. These steps achieve high-precision detection and quantitative evaluation of wall defects through precise signal processing and analysis.

[0026] In some embodiments, the wavelet filtering process in step 4 satisfies the signal-to-noise ratio improvement relationship: in is the signal-to-noise ratio of the preprocessed signal, is the signal-to-noise ratio of the denoised signal.

[0027] It should be noted that the wavelet filtering mentioned in the present invention can significantly improve the signal-to-noise ratio of the signal. The specific improvement relationship is that the output signal-to-noise ratio is equal to the input signal-to-noise ratio plus 18.81 decibels. The signal-to-noise ratio (SNR) is an important parameter in the field of signal processing. It represents the ratio of signal power to noise power and is usually measured in decibels (dB). In ultrasonic testing, a higher signal-to-noise ratio means that the effective information in the signal is more prominent and the noise interference is smaller, thereby enabling more accurate detection of defects inside the wall. Through wavelet filtering, the noise component in the original received signal can be effectively removed, the quality and reliability of the signal can be enhanced, and a clearer signal basis can be provided for subsequent defect detection and analysis.

[0028] Specifically, wavelet filtering is a signal processing method based on wavelet transform. Wavelet transform is a technology that decomposes a signal into different frequency components, which can convert the signal from the time domain to the wavelet domain. In the wavelet domain, the effective components and noise components of the signal can be separated. By removing the noise coefficient and then performing an inverse wavelet transform on the effective signal coefficient, the denoised signal can be reconstructed. The input signal-to-noise ratio here refers to the signal-to-noise ratio of the signal after preprocessing by the receiving drive circuit, and the output signal-to-noise ratio refers to the signal-to-noise ratio of the signal after wavelet filtering. The improvement of 18.81 decibels shows that the denoising effect of wavelet filtering on the signal is very significant, which can greatly enhance the effective information of the signal and improve the accuracy of detection.

[0029] Preferably, the specific implementation process of wavelet filtering is as follows: First, a wavelet transform is performed on the signal preprocessed by the receiving drive circuit. The appropriate wavelet basis function and decomposition level are selected to decompose the signal into wavelet coefficients of different frequency components. Then, by setting a threshold function, the noise components in the wavelet coefficients are separated and removed. The selection and setting of the threshold function is crucial to the denoising effect and can be optimized based on the characteristics of the signal and the distribution of the noise. Finally, an inverse wavelet transform is performed on the effective signal coefficients after removing the noise coefficients to reconstruct the denoised signal. This process not only effectively removes noise but also preserves important features in the signal, providing high-quality signal input for subsequent time-frequency joint analysis and defect detection.

[0030] In some embodiments, the semantic image segmentation operation in step 6 includes the following sub-steps: Sub-step 6.1: Perform a padding operation on the input time-frequency domain feature signal to obtain a padded feature map; Sub-step 6.2: Perform convolution operations on the padded feature map to extract vertical and horizontal features; Sub-step 6.3: Perform a maximum pooling operation on the feature map output by the convolution operation to obtain compressed feature data; Sub-step 6.4: Output the defect area segmentation map.

[0031] It should be noted that the semantic image segmentation operation mentioned in the present invention is implemented based on a multi-scale convolutional neural network (CNN). Its purpose is to process the time-frequency domain feature signals and output a defect area segmentation map with the same resolution as the input image. Semantic image segmentation is an image processing technology that not only identifies targets in an image but also classifies each pixel in the image and labels the category to which it belongs. In the present invention, after the input time-frequency domain feature signals are processed, the output segmentation map can clearly identify the location and range of wall defects, providing intuitive visualization results for the detection and analysis of wall defects.

[0032] Specifically, the semantic image segmentation operation includes the following sub-steps: first, a padding operation is performed on the input time-frequency domain feature signal to obtain a padded feature map. The padding operation is to ensure that the boundary information of the image is not lost during the convolution operation; then a convolution operation is performed on the padded feature map to extract vertical and horizontal features. The convolution operation is to calculate a new feature map by sliding the convolution kernel on the image, which is used to extract local features in the image; then a maximum pooling operation is performed on the feature map output by the convolution operation to obtain compressed feature data. The maximum pooling operation takes the maximum value in the local area, which is used to reduce the spatial size of the feature map while retaining the most important features; finally, a defect area segmentation map is output. The input image here refers to the feature signal after the time-frequency joint analysis, and the output segmentation map is the image with the defect area marked after CNN processing.

[0033] Preferably, the construction process of the multi-scale convolutional neural network is as follows: the network receives a time-frequency domain feature signal of size N×N as input, and extracts the features of the image through multi-layer convolution operations. Each layer of convolution operation uses different convolution kernels to extract features of different scales, thereby realizing multi-scale analysis. After the convolution operation, the size of the feature map is further compressed by the maximum pooling operation, reducing the amount of calculation while retaining key features. After multiple layers of convolution and pooling operations, the network outputs a segmentation map with the same resolution as the input image, in which each pixel is labeled with its corresponding category, such as wall defects or non-defective areas. In order to improve the accuracy of segmentation, the network will use a large amount of labeled data for learning during the training process to optimize the parameters of the convolution kernel so that the network can better identify and segment wall defects.

[0034] In some embodiments, the method further includes step nine: Perform spatiotemporal alignment on the RGB image V and the infrared thermal image T, and optimize the modal weight coefficients through non-negative matrix decomposition , establish a multimodal fusion model: in, represents the tensor product, is the fused feature output, is an RGB image, For infrared thermal imaging, is the modal weight coefficient, is the number of modes, The first modal data, For infrared thermal imaging modal data.

[0035] It should be noted that the multimodal fusion model mentioned in the present invention is constructed by performing spatiotemporal alignment operations on RGB images and infrared thermal images, and optimizing the modal weight coefficients using non-negative matrix decomposition. Multimodal data fusion refers to the integration of data from different sensors or different physical modalities to obtain more comprehensive and accurate information. In the present invention, RGB images provide visible light image information of the wall, while infrared thermal images reflect the heat distribution of the wall. Through the spatiotemporal alignment operation, it can be ensured that the data of the two modalities are consistent in time and space, thereby providing a basis for subsequent fusion processing. Non-negative matrix decomposition NMF is a commonly used matrix decomposition method, which can decompose the original data into weight coefficients of multiple modalities. Optimizing these weight coefficients can better reflect the importance and contribution of different modal data, thereby improving the quality of the fusion results.

[0036] Specifically, the spatiotemporal alignment operation refers to matching the RGB image and the infrared thermal image in time and space so that the data of the two modalities can correspond to the same wall area and the same time point. The process of optimizing the modal weight coefficient by non-negative matrix decomposition is achieved by expressing the fused feature output as the weighted sum of each modal data. Among them, the modal weight coefficient represents the importance of each modal data in the fusion result. By optimizing these coefficients, the fused feature output can be closer to the actual wall defect situation. The tensor product is a mathematical operation used to combine data of different modalities to generate a fused feature output. In the present invention, the RGB image and the infrared thermal image are respectively used as the data of the two modalities, and are fused through the modal weight coefficients optimized by tensor product and non-negative matrix decomposition, so as to obtain more comprehensive and accurate wall defect information.

[0037] Preferably, the steps for constructing the multimodal fusion model are as follows: First, collect the RGB image and infrared thermal image data of the wall. Then, perform spatiotemporal alignment operations on the data of these two modalities to ensure that they are consistent in time and space. Next, optimize the modal weight coefficients using non-negative matrix decomposition. Specifically, the fused feature output is represented as the weighted sum of the RGB image and the infrared thermal image, where the weight coefficients are obtained by non-negative matrix decomposition. Finally, the optimized modal weight coefficients are combined with the corresponding modal data through tensor product to generate the fused feature output. In practical applications, the parameters of the non-negative matrix decomposition and the optimization algorithm of the modal weight coefficients can be adjusted according to the needs of wall defect detection to improve the accuracy and reliability of the fusion results.

[0038] In some embodiments, the ultrasonic wave propagation in step 1 satisfies the wave equation: in, is the density of the wall material being measured, is the particle displacement, is the elastic coefficient, For time, is the spatial coordinate.

[0039] It should be noted that the ultrasonic propagation mentioned in this invention satisfies the wave equation, a fundamental physical law describing the propagation of ultrasonic waves in a medium. The wave equation is a partial differential equation that characterizes the relationship between the displacement of a particle in a medium and its relationship to time and space. In this invention, parameters such as the density, particle displacement, and elastic modulus of the wall material being tested jointly determine the propagation characteristics of ultrasonic waves in the wall. By solving the wave equation, the propagation path, reflection, and transmission of ultrasonic waves in the wall can be predicted, thus providing a theoretical basis for detecting wall defects.

[0040] Specifically, the parameters in the wave equation have clear physical meanings. The density ρ of the wall material being measured refers to the mass per unit volume of the wall material. It reflects the material's compactness and has a direct impact on the propagation speed and attenuation of ultrasonic waves. The particle displacement u represents the change in position of a particle in the medium under the action of ultrasonic waves and is a key variable describing energy transfer during ultrasonic propagation. The elastic modulus C is a physical property of the material, characterizing the material's ability to deform when subjected to external forces. It determines the propagation speed and energy distribution of ultrasonic waves in the medium. Time t and spatial coordinates (x, y, z) describe the time and position information during ultrasonic propagation. Through the relationship between these parameters, the wave equation accurately describes the propagation behavior of ultrasonic waves in wall media, including phenomena such as reflection, refraction, and attenuation.

[0041] Preferably, the propagation of ultrasonic waves in a wall can be analyzed using the following steps: First, determine the density ρ and elastic modulus C of the wall material being tested based on its properties. These parameters can be obtained through experimental measurement or by consulting a material manual. These parameters are then substituted into the wave equation and, combined with specific boundary conditions, such as those on the wall surface and at the boundaries of internal defects, the equation is solved to determine the ultrasonic wave's propagation path and reflected signal within the wall. In practical applications, the wave equation can be solved using numerical simulation methods, such as finite element analysis, with input parameters including the wall's geometric dimensions, material properties, and the ultrasonic wave's transmission frequency. This simulation allows prediction of ultrasonic wave propagation within the wall, providing theoretical support for defect detection. For example, when an ultrasonic wave encounters a defect within a wall, reflection and transmission occur. By analyzing the intensity and time delay of the reflected signal, the defect's location and size can be inferred.

[0042] In some embodiments, the time reversal detection process in step 7 satisfies: in, is the input time-frequency domain characteristic signal, is the time-reversed signal output, For time, is the attenuation compensation factor under air coupling conditions.

[0043] It should be noted that the time-reversal detection processing mentioned in this invention is a processing method based on the time-reversal symmetry of the signal, which is used to enhance the signal characteristics and improve the accuracy of defect detection. Time-reversal detection achieves signal enhancement and focusing by reversing the input signal on the time axis. This method is particularly suitable for ultrasonic detection, because ultrasonic signals are attenuated and scattered by the medium during propagation, resulting in weakened signal strength and blurred characteristics. Through time-reversal detection processing, the signal characteristics can be effectively restored, improving the sensitivity and accuracy of detection.

[0044] Specifically, the input signal in the time-reversal detection process refers to the time-frequency domain characteristic signal after preprocessing and wavelet filtering. The time-reversal signal output is obtained by performing a time-reversal operation on the input signal, that is, reversing the order of the signal on the time axis. The attenuation compensation factor β under air coupling conditions is an important parameter, which is used to compensate for the energy attenuation of the signal when it propagates in the air medium. In ultrasonic testing, due to the large difference in acoustic impedance between air and wall materials, the attenuation of the signal in air is more significant. Therefore, it is necessary to adjust the signal strength through the attenuation compensation factor to ensure that the propagation characteristics of the signal in the wall can accurately reflect the defect information. The output signal of the time-reversal detection process can be used for further defect analysis and positioning.

[0045] Preferably, the implementation steps of the time-reversal detection process are as follows: First, obtain the time-frequency domain characteristic signal after wavelet filtering, which has removed the noise component and retained the effective characteristics. Then, perform a time-reversal operation on the signal, that is, reverse the signal on the time axis to obtain a time-reversed signal. Next, based on the attenuation characteristics under air-coupled conditions, introduce an attenuation compensation factor β to adjust the time-reversed signal. Specifically, the attenuation compensation factor can be obtained through experimental measurement or theoretical calculation, and it reflects the degree of energy loss when the signal propagates in the air medium. By compensating for attenuation, the strength of the signal can be enhanced, making it closer to the actual situation when propagating in the wall. Finally, the time-reversed signal is used for defect detection, for example, by analyzing the reflection intensity and time delay of the signal to determine the location and size of the internal defect of the wall. This method can effectively improve the resolution and reliability of detection and is particularly suitable for complex environments in air-coupled ultrasonic detection.

[0046] In some embodiments, the driving circuit of the air-coupled ultrasonic detection device includes: Oscillation source module, used to generate high-frequency square wave signal; The frequency division circuit module uses a dual D flip-flop CD4013 to perform frequency division processing on the high-frequency square wave signal; The power amplifier module uses the IRF830 amplifier to amplify the power of the divided signal; The non-polarized capacitive coupling module couples the power-amplified signal to the transmitting probe.

[0047] It should be noted that the drive circuit of the air-coupled ultrasonic detection device mentioned in this invention is a key component for achieving ultrasonic signal transmission. The drive circuit generates and amplifies high-frequency signals, ultimately driving the ultrasonic transmitting probe to emit ultrasonic pulses. The oscillation source module is responsible for generating high-frequency square wave signals, the frequency divider circuit module is used to adjust the signal frequency, the power amplifier module amplifies the signal power, and the non-polarized capacitive coupling module couples the signal to the transmitting probe. These modules work together to ensure that the ultrasonic signal can effectively propagate through the air medium and penetrate the wall being tested.

[0048] Specifically, the oscillation source module is the core part of the drive circuit. It generates a high-frequency square wave signal through components such as a crystal oscillator, providing the basic signal for subsequent circuits. The frequency division circuit module uses a dual D flip-flop CD4013 to divide the high-frequency square wave signal and adjust the signal frequency to adapt to the operating frequency of the ultrasonic transmitting probe. The power amplifier module uses an IRF830 amplifier to amplify the power of the divided signal to ensure that the signal has sufficient energy to penetrate the air and walls. The non-polarized capacitive coupling module is used to couple the amplified signal to the transmitting probe without distortion, ensuring signal integrity and stability. The parameter settings of these modules need to be optimized according to the specific needs of ultrasonic testing, such as the frequency of the oscillation source, the frequency division ratio, the gain of the amplifier, etc., to ensure that the ultrasonic signal can be effectively transmitted under air coupling conditions.

[0049] Preferably, the driving circuit construction steps of the air-coupled ultrasonic detection device are as follows: First, select a suitable oscillation source module, such as a ZOOM crystal oscillator, to generate a stable high-frequency square wave signal. Then, the high-frequency signal is divided by a frequency division circuit module, such as a dual D flip-flop CD4013, and the signal frequency is adjusted to 1MHz, which is suitable for the operation of the ultrasonic probe. Next, a power amplifier module, such as an IRF830 amplifier, is used to amplify the power of the divided signal to ensure that the signal has sufficient energy to penetrate the air and the wall. Finally, the amplified signal is coupled to the transmitting probe through a non-polarized capacitor coupling module. In actual applications, the frequency of the oscillation source and the gain of the power amplifier can be adjusted according to factors such as the characteristics of the wall material and the detection distance to optimize the signal propagation effect. For example, for thicker walls, higher signal power is required to ensure that the signal can penetrate the wall and be detected by the receiving probe.

[0050] In some embodiments, the defect quantitative analysis operation in step eight includes: Sub-step 8.1: performing a C-scan image reconstruction operation on the defect region segmentation map to output a defect spatial distribution map; Sub-step 8.2: Calculating the transmitted wave group velocity based on the time-reversed signal; Sub-step 8.3: Invert the exponential relationship between the surface crack depth d and the internal hollow volume V based on the transmitted wave group velocity: , where V is the internal hollow volume; k is the material constant, which is a fixed coefficient related to the wall material properties; c is the material constant, which reflects the rate at which the internal hollow volume changes with the depth of the surface crack; d is the depth of the surface crack; Sub-step 8.4: Mark the defect area with a diameter of 5 mm in the defect spatial distribution map.

[0051] It should be noted that the defect quantification analysis process in this invention combines image processing and signal analysis to accurately quantify wall defects. The defect region segmentation map, generated using image segmentation technology, clearly identifies the location and extent of the wall defect. The time-reversed signal is used to analyze the propagation characteristics of ultrasonic waves in the wall, thereby inferring the severity of the defect. This method not only detects the presence of wall defects but also quantifies their size and depth, providing a scientific basis for subsequent repair and maintenance.

[0052] Specifically, the quantitative defect analysis operation includes the following key steps: First, a C-scan image reconstruction operation is performed on the defect area segmentation map. C-scan is an ultrasonic imaging technology that generates a spatial distribution map of defects by performing a two-dimensional scan of the wall surface and recording the reflected signals. Second, the transmission wave group velocity is calculated based on the time-reversed signal. The transmission wave group velocity refers to the speed at which ultrasonic waves propagate through the wall material, which is related to the density and elastic properties of the wall material. Finally, the relationship between the surface crack depth and the internal hollow volume is inverted based on the transmission wave group velocity. This relationship is obtained by fitting experimental data, where the surface crack depth refers to the depth of the cracks on the wall surface, and the internal hollow volume refers to the volume of the hollow portion inside the wall. The determination of these parameters requires analysis in conjunction with the physical properties of the wall material.

[0053] Preferably, the specific implementation steps of the defect quantitative analysis are as follows: First, the time-frequency domain characteristic signals are processed using a multi-scale convolutional neural network to generate a defect area segmentation map. Then, the segmentation map is reconstructed by C-scan image to obtain a spatial distribution map of the defects, clarifying the location and range of the defects. Next, the transmission wave group velocity is calculated based on the time-reversed signal. This process requires spectral analysis of the time-reversed signal, and by analyzing the frequency components of the signal, the propagation speed of the ultrasonic wave in the wall is inferred. Finally, the relationship between the surface crack depth and the internal hollow volume is inverted based on the transmission wave group velocity. Specifically, by experimentally measuring the transmission wave group velocity corresponding to cracks of different depths, the relevant characteristic parameters of the wall material are fitted, and then these parameters are used to calculate the internal hollow volume. In practical applications, the accuracy of the defect quantitative analysis can be further improved by adjusting the resolution of the C-scan and the processing accuracy of the time-reversed signal.

[0054] In some embodiments, the propagation of ultrasonic waves in the solid medium in step 1 satisfies the isotropic wave equation: in, is the density of the wall material being measured, u is the particle displacement, t is the time, x and y are the spatial coordinates, and is the Mellar constant.

[0055] It should be noted that the propagation of ultrasound in solid media mentioned in this invention satisfies the isotropic wave equation, which describes the propagation characteristics of ultrasound in a uniform and isotropic solid medium. The isotropic wave equation reflects the relationship between particle displacement and time and space coordinates when ultrasound propagates in a medium. This equation can better understand the propagation laws of ultrasound in solid media such as walls, providing theoretical support for wall defect detection. This equation takes into account factors such as the density of the medium, particle displacement, and elastic modulus, which together determine the propagation speed and attenuation characteristics of ultrasound.

[0056] Specifically, the parameters in the isotropic wave equation have clear physical meanings. The density ρ of the wall material being measured refers to the mass per unit volume of the wall material. It reflects the material's compactness and has a direct impact on the propagation speed and attenuation of ultrasonic waves. The particle displacement u represents the change in position of a particle in the medium under the action of ultrasonic waves and is a key variable describing energy transfer during ultrasonic propagation. Time t and spatial coordinates (x, y) represent the time and position information of ultrasonic propagation, respectively. The elastic coefficients λ and μ are physical properties of the material, characterizing the material's ability to deform when subjected to external forces. λ and μ are the Lame constants, respectively, which determine the propagation speed and energy distribution of ultrasonic waves in the medium. Through the relationship between these parameters, the isotropic wave equation accurately describes the propagation behavior of ultrasonic waves in wall media, including phenomena such as reflection, refraction, and attenuation.

[0057] Preferably, the propagation of ultrasonic waves in a wall can be analyzed using the following steps: First, determine the density ρ and elastic moduli λ and μ based on the properties of the wall material being tested. These parameters can be obtained through experimental measurement or by consulting a material manual. These parameters are then substituted into the isotropic wave equation and, combined with specific boundary conditions, such as those on the wall surface and at the boundaries of internal defects, the equation is solved to determine the ultrasonic wave propagation path and reflected signal within the wall. In practical applications, the wave equation can be solved using numerical simulation methods, such as finite element analysis, with input parameters including the wall's geometric dimensions, material properties, and the ultrasonic wave's transmission frequency. This simulation allows prediction of ultrasonic wave propagation within the wall, providing theoretical support for defect detection. For example, when an ultrasonic wave encounters a defect within a wall, reflection and transmission occur. By analyzing the intensity and time delay of the reflected signal, the defect's location and size can be inferred.

[0058] In some embodiments, the signal preprocessing in step 3 further includes a low-pass filtering operation, and the sound field equation when there is no external force is: And the signal enhancement processing satisfies wavelet reconstruction: in, is the density of the wall material being measured, u is the displacement of the mass point, t is the time, C is the elastic coefficient, is the Laplace operator, For the enhanced signal output, is the wavelet transform coefficient, a is the scale parameter, b is the translation parameter, is the wavelet basis function.

[0059] It should be noted that the signal preprocessing mentioned in the present invention also includes a low-pass filtering operation, which is to further remove high-frequency noise and ensure the purity of the signal. Low-pass filtering is a common signal processing technology that allows low-frequency signals to pass through while attenuating or suppressing high-frequency signals. In ultrasonic detection, high-frequency noise can interfere with signal analysis, leading to misjudgment or decreased detection accuracy. Through low-pass filtering operations, these high-frequency interferences can be effectively removed and the quality of the signal can be improved. In addition, the signal enhancement processing satisfies wavelet reconstruction, which is a method of enhancing the signal using wavelet transform. By reconstructing the wavelet coefficients of the signal, the effective information in the signal can be further highlighted and the detectability of the signal can be improved.

[0060] Specifically, the parameters involved in the low-pass filtering operation include the cutoff frequency and the order of the filter. The cutoff frequency refers to the frequency point at which the filter begins to attenuate the signal, and is usually set according to the characteristics of the signal and the frequency range of the noise. The order of the filter determines the steepness of the filter. The higher the order, the narrower the transition band of the filter, but it may introduce more phase distortion. In signal enhancement processing, wavelet reconstruction is to reconstruct the signal by performing a wavelet transform on the signal and then selecting the appropriate wavelet basis function and reconstruction algorithm. Wavelet transform coefficients, scale parameters, and translation parameters are key variables in wavelet transform, which determine the representation of the signal at different frequency and time scales. By adjusting these parameters, the signal enhancement effect can be optimized.

[0061] Preferably, the specific implementation steps of signal preprocessing and enhancement are as follows: First, determine the cutoff frequency of the low-pass filter, usually selecting a value slightly higher than the main frequency component of the signal to ensure that the low-frequency part of the signal is not affected while effectively removing high-frequency noise. Then, select the appropriate filter order according to actual needs to balance the filtering effect and phase distortion. Next, perform a wavelet transform on the signal after low-pass filtering, and select appropriate wavelet basis functions, such as Daubechies or Morlet wavelets. These basis functions can effectively decompose the different frequency components of the signal. In the wavelet domain, by adjusting the scale parameter and translation parameter, the wavelet coefficients of the signal are processed to enhance the effective components of the signal. Finally, reconstruct the signal through the inverse wavelet transform to obtain the enhanced signal output. In practical applications, these parameters can be optimized through experiments or simulations to achieve the best signal processing effect.

[0062] The above embodiments of the present invention have the following beneficial effects: 1. By adopting air-coupled ultrasonic detection technology, the problem of reduced coating anti-corrosion performance caused by traditional coupling agent residue is effectively solved, the corrosion effect of coupling agent residue on the substrate surface is avoided, and the risk of reduced coating adhesion caused by incomplete coupling agent cleaning is eliminated. It is particularly suitable for building structure inspection scenarios that require long-term anti-corrosion protection.

[0063] 2. The innovative signal processing solution combines wavelet filtering and time-frequency joint analysis methods to significantly improve the sensitivity and accuracy of defect detection. It can effectively identify tiny defect characteristics that are difficult to detect with traditional methods. It solves the technical difficulties of severe signal attenuation and low signal-to-noise ratio in conventional ultrasonic detection in complex building structures, and provides more reliable detection data for building structure health assessment.

[0064] 3. The intelligent analysis system based on multi-scale convolutional neural networks and time reversal technology realizes the automatic identification and quantitative evaluation of defects, overcoming the subjectivity and instability of manual interpretation. At the same time, by establishing a correlation model between hollow volume and crack depth, it provides a scientific basis for damage assessment and maintenance decision-making of building structures, and solves the pain point of insufficient quantitative analysis of traditional detection methods.

[0065] Furthermore, the storage medium of the embodiment of the present application stores program instructions that can implement all the above methods, wherein the program instructions can be stored in the above storage medium in the form of a software product, including a number of instructions for causing a computer device (which can be a personal computer, server, or network device, etc.) or a processor to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, or a terminal device such as a computer, server, mobile phone, or tablet.

[0066] The above descriptions merely illustrate some preferred embodiments of the present invention and the underlying technical principles. Those skilled in the art should understand that the scope of the invention encompassed by the embodiments of the present invention is not limited to technical solutions formed by specific combinations of the aforementioned technical features. It also encompasses other technical solutions formed by any combination of the aforementioned technical features or their equivalents, without departing from the aforementioned inventive concept. For example, a technical solution formed by replacing the aforementioned features with (but not limited to) technical features with similar functions disclosed in the embodiments of the present invention.

Claims

1. A method for ultrasonic time-frequency joint analysis under uncoupled conditions, characterized in that: It includes the following steps, performed in order: Step 1: Use an air-coupled ultrasonic detection device to transmit an ultrasonic pulse signal with a frequency of 1 MHz to the wall under test. The air-coupled ultrasonic detection device uses air as a coupling medium, and the transmitted ultrasonic wave penetrates the wall under test; Step 2: Receive the reflected ultrasonic signal through the receiving probe of the air-coupled ultrasonic detection device to obtain the original received signal; Step 3: Inputting the original received signal into the receiving driving circuit for signal preprocessing, wherein the signal preprocessing includes signal amplification, signal shaping and detection filtering, and outputting the preprocessed signal; Step 4: Perform wavelet filtering on the preprocessed signal to obtain a denoised signal, which includes: Sub-step 4.1: Perform wavelet transform on the preprocessed signal to obtain a wavelet domain signal; Sub-step 4.2: Separate the effective signal coefficient and noise coefficient in the wavelet domain signal; Sub-step 4.3: discard the noise coefficient; Sub-step 4.4: Perform inverse wavelet transform on the effective signal coefficients to reconstruct the denoised signal; Step 5: Perform a time-frequency joint analysis operation on the denoised signal to extract the time-frequency domain feature signal; Step 6: Input the time-frequency domain feature signal into a multi-scale convolutional neural network to perform a semantic image segmentation operation, and output a defect area segmentation map, wherein the input image and the output segmentation map of the multi-scale convolutional neural network have the same resolution; Step 7: Perform time reversal detection processing on the time-frequency domain feature signal to obtain a time reversal signal; Step 8: Generate quantitative detection results of wall defects based on the defect area segmentation map and time reversal signal.

2. The ultrasonic time-frequency joint analysis method for uncoupled conditions according to claim 1, characterized in that: The wavelet filtering process described in step 4 satisfies the signal-to-noise ratio improvement relationship: in is the signal-to-noise ratio of the preprocessed signal, is the signal-to-noise ratio of the denoised signal.

3. The ultrasonic time-frequency joint analysis method for uncoupled conditions according to claim 1, characterized in that: The semantic image segmentation operation in step 6 includes the following sub-steps: Sub-step 6.1: Perform a padding operation on the input time-frequency domain feature signal to obtain a padded feature map; Sub-step 6.2: Perform convolution operations on the padded feature map to extract vertical and horizontal features; Sub-step 6.3: Perform a maximum pooling operation on the feature map output by the convolution operation to obtain compressed feature data; Sub-step 6.4: Output the defect area segmentation map.

4. The ultrasonic time-frequency joint analysis method for uncoupled conditions according to claim 1, characterized in that: Also includes step nine: Perform spatiotemporal alignment on the RGB image V and the infrared thermal image T, and optimize the modal weight coefficients through non-negative matrix decomposition , establish a multimodal fusion model: in, represents the tensor product, is the fused feature output, is an RGB image, For infrared thermal imaging, is the modal weight coefficient, is the number of modes, The first modal data, For infrared thermal imaging modal data.

5. The ultrasonic time-frequency joint analysis method for uncoupled conditions according to claim 1, characterized in that: In step 1, ultrasonic wave propagation satisfies the wave equation: in, is the density of the wall material being measured, is the particle displacement, is the elastic coefficient, For time, is the spatial coordinate.

6. The ultrasonic time-frequency joint analysis method for uncoupled conditions according to claim 1, characterized in that: The time reversal detection process in step 7 satisfies: in, is the input time-frequency domain characteristic signal, is the time-reversed signal output, For time, is the attenuation compensation factor under air coupling conditions.

7. The ultrasonic time-frequency joint analysis method for uncoupled conditions according to claim 1, characterized in that: The driving circuit of the air-coupled ultrasonic detection device includes: Oscillation source module, used to generate high-frequency square wave signal; The frequency division circuit module uses a dual D flip-flop CD4013 to perform frequency division processing on the high-frequency square wave signal; The power amplifier module uses the IRF830 amplifier to amplify the power of the divided signal; The non-polarized capacitive coupling module couples the power-amplified signal to the transmitting probe.

8. The ultrasonic time-frequency joint analysis method for uncoupled conditions according to claim 1, characterized in that: The defect quantitative analysis operation in step eight includes: Sub-step 8.1: performing a C-scan image reconstruction operation on the defect region segmentation map to output a defect spatial distribution map; Sub-step 8.2: Calculating the transmitted wave group velocity based on the time-reversed signal; Sub-step 8.3: Invert the exponential relationship between the surface crack depth d and the internal hollow volume V based on the transmitted wave group velocity: , where V is the internal hollow volume; k is the material constant, which is a fixed coefficient related to the wall material properties; c is the material constant, which reflects the rate at which the internal hollow volume changes with the depth of the surface crack; d is the depth of the surface crack; Sub-step 8.4: Mark the defect area with a diameter of 5 mm in the defect spatial distribution map.

9. The ultrasonic time-frequency joint analysis method for uncoupled conditions according to claim 1, characterized in that: In step 1, the propagation of ultrasonic waves in solid media satisfies the isotropic wave equation: in, is the density of the wall material being measured, u is the particle displacement, t is the time, x and y are the spatial coordinates, and is the Mellar constant.

10. The ultrasonic time-frequency joint analysis method for uncoupled conditions according to claim 1, characterized in that: The signal preprocessing in step 3 also includes a low-pass filtering operation. When there is no external force, the sound field equation is: And the signal enhancement processing satisfies wavelet reconstruction: in, is the density of the wall material being measured, u is the displacement of the mass point, t is the time, C is the elastic coefficient, is the Laplace operator, For the enhanced signal output, is the wavelet transform coefficient, a is the scale parameter, b is the translation parameter, is the wavelet basis function.

Images