Multi-modal mechanical wave detection method and system based on boundary rebound characteristics

Abstract

The invention discloses a multi-modal mechanical wave detection method and system based on boundary rebound characteristics, and relates to the technical field of nondestructive testing, and the method comprises the steps: collecting the waveform, frequency and amplitude of a multi-modal mechanical wave through a sensor, and carrying out the preprocessing; analyzing reflection characteristics of the multi-modal mechanical waves encountering boundaries when the multi-modal mechanical waves are propagated in the target object, calculating boundary rebound values of the multi-modal mechanical waves, and selecting the mechanical wave with the highest boundary rebound value as a depth waveform characteristic; and constructing a relation model, inputting the depth waveform features into the relation model, predicting the severity of the defect, and generating an early warning scheme. According to the method, noise interference is effectively removed, the signal definition is improved, a foundation is laid for accurate analysis of boundary rebound characteristics of mechanical waves, powerful data support is provided for subsequent defect evaluation, quantitative evaluation and real-time monitoring of internal corrosion defects of a target object are achieved, and the method is suitable for popularization and application. And the practicability and the engineering application value of the detection system are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of nondestructive testing, and in particular to a multi-modal mechanical wave testing method and system based on boundary rebound characteristics. Background Art

[0002] In recent years, mechanical wave detection technology has been widely used in nondestructive testing, structural health monitoring and material characterization. Mechanical waves, including sound waves, elastic waves and ultrasonic waves, have attracted much attention because they can penetrate the interior of materials and respond sensitively to defects, internal stresses and structural changes. In practical applications, mechanical waves of different modes can provide rich detection information due to their different propagation characteristics and interactions with material boundaries. In particular, multimodal mechanical wave detection technology, by integrating the characteristics of multiple wave modes, provides the possibility for accurate detection of complex structures and heterogeneous materials. However, with the increase in application demand and the improvement of detection accuracy, in-depth research on the propagation characteristics of mechanical waves inside target objects and how to extract key waveform features have become hot topics in current research. Current research focuses on waveform signal processing, propagation path simulation and multimodal data fusion, which lays a theoretical foundation and practical basis for the further development of mechanical wave detection technology.

[0003] Although multimodal mechanical wave detection technology has significant advantages in complex media environments, existing methods still have some limitations and shortcomings. Existing technologies usually focus on single-mode waves and lack comprehensive analysis of multimodal wave characteristics. This limitation leads to the loss of some important information in the detection process, especially the waveform change characteristics in the boundary reflection process cannot be fully utilized. When the mechanical wave interacts with the boundary of the target object, the characteristics of the reflected wave can reflect the boundary characteristics and internal structure of the material, but the existing technology's research on this characteristic is relatively superficial, and a systematic analysis method has not yet been formed. In the detection and optimization selection of multimodal waves, there is a lack of a scientific and effective quantitative indicator, which makes it difficult to select the best wave mode with high detection sensitivity in a complex detection environment, resulting in limited robustness and accuracy of the detection results. Therefore, improving the accuracy and practicality of mechanical wave detection is an urgent problem to be solved in the current technological development. Summary of the invention

[0004] In view of the above-mentioned problems, the present invention is proposed.

[0005] Therefore, the technical problems solved by the present invention are: the existing multimodal mechanical wave detection technology has the problem of being unable to effectively utilize the boundary rebound characteristics for analysis, the lack of effective screening of waveforms sensitive to defects, and how to select the most representative deep waveform features from multiple modal mechanical waves to achieve accurate detection and evaluation of internal defects.

[0006] To solve the above technical problems, the present invention provides the following technical solutions: a multimodal mechanical wave detection method based on boundary rebound characteristics, comprising using sensors to collect the waveforms, frequencies and amplitudes of multiple modal mechanical waves and preprocessing them; analyzing the reflection characteristics of multiple modal mechanical waves when encountering boundaries when propagating inside the target object, calculating the boundary rebound values ​​of the multiple modal mechanical waves, and selecting the mechanical wave with the highest boundary rebound value as the deep waveform feature; constructing a relationship model, inputting the deep waveform features into the relationship model, predicting the severity of the defect, and generating an early warning plan.

[0007] As a preferred solution of the multimodal mechanical wave detection method based on boundary rebound characteristics described in the present invention, the use of sensors to collect the waveforms, frequencies and amplitudes of multiple modal mechanical waves includes selecting MEMS sensors, configuring the sampling frequency and sampling time of the sensors, setting the time window for data acquisition, and collecting signal data of longitudinal waves, transverse waves, surface waves and plate waves when they propagate inside the target object.

[0008] As a preferred solution of the multi-modal mechanical wave detection method based on boundary rebound characteristics described in the present invention, the pre-processing process includes using a filtering technology Butterworth filter to eliminate the noise interference of multiple modal mechanical waves to obtain a denoised signal x filtered (t), normalize the signal so that all signal values ​​fall within the same range, and calculate the normalized value x norm (t), expressed as:

[0009]

[0010] Among them, x min is the minimum value of the denoised signal, x max is the maximum value of the denoised signal, x norm (t) is the normalized value.

[0011] As a preferred solution of the multimodal mechanical wave detection method based on boundary rebound characteristics described in the present invention, the analysis of the reflection behavior of multiple modal mechanical waves when encountering a boundary while propagating inside a target object includes applying multimodal mechanical waves at different positions of the target object, and using a high-speed camera and a laser Doppler vibrometer to record the response intensity of the waveform, using finite element analysis software COMSOL Multiphysics to simulate the response intensity of multiple modal waves at corrosion defects, and in the absence of corrosion defects, collecting background noise signals Q, and calculating the standard deviation σ of the background noise as an indicator of the noise level.

[0012] The reflection behavior of the waveform at different boundaries is recorded, and a multi-scale analysis method is introduced to analyze the boundary rebound characteristics at different scales from macro to micro, including using wavelet transform to remove noise and retain the main features of the waveform in the real reflection characteristics. The reflected wave field intensity S(t,s) after denoising is calculated and expressed as:

[0013]

[0014] Among them, S(t,s) is the reflected wave intensity after denoising, ψ is the wavelet basis function, s is the scale parameter, t is the time parameter, x norm (τ) is the original signal, is an integral operation, dτ is an infinitesimal increment of the time variable τ, the signal data of the waveform characteristics are normalized to the same order of magnitude, and the normalized signal S is calculated norm (t,s), expressed as:

[0015]

[0016] Among them, S min is the minimum value of the signal after wavelet transformation, S max is the maximum value of the signal after wavelet transformation, S norm (t,s) is the normalized signal.

[0017] Collect the background noise signal Q and calculate the standard deviation σ of the background noise as an indicator of the noise level, including calculating the mean μ and the variance σ 2 , calculate the standard deviation σ, and evaluate the level of background noise based on the calculated σ. When σ exceeds the preset threshold σ1, it means that the background noise level is too high. Re-collect the background noise signal and use the adaptive filter to further denoise it. Calculate the denoised background noise signal W(t), which is expressed as:

[0018] W(t)=QH adapt (J)·E(t)

[0019] Where W(t) is the background noise signal after denoising, H adapt (J) is the transfer function of the adaptive filter, E(t) is the estimated noise signal, and the standard deviation σ2 of the denoised background noise signal is recalculated as:

[0020]

[0021] Among them, σ 2 is the standard deviation of the background noise signal after denoising, μ clean is the average value of the background noise signal after denoising, (W(t i )-μ clean ) 2is the square of the difference between each denoised background noise signal value and the average value, is the summation symbol, W(t i ) is the value of the denoised background noise signal at the i-th time point, For the averaging operation, N is the length of the signal and i is each time point.

[0022] As a preferred solution of the multimodal mechanical wave detection method based on boundary rebound characteristics described in the present invention, wherein: the selection of the mechanical wave with the highest boundary rebound value as the depth waveform feature includes setting the boundary rebound value to R(t,s), expressed as:

[0023]

[0024] Among them, R(t,s) is the boundary rebound value, S norm (t, s) is the normalized reflected wave field strength. The mechanical wave with the highest boundary rebound value is selected as the depth waveform feature. The depth waveform feature F is calculated. depth , expressed as:

[0025]

[0026] Among them, F depth is the deep waveform feature, It means finding the values ​​of t and s that maximize R(t,s) over all times and scales.

[0027] As a preferred solution of the multimodal mechanical wave detection method based on boundary rebound characteristics described in the present invention, wherein: the relationship model is constructed according to the corrosion defect and the depth waveform feature, the depth waveform feature is input into the relationship model, and the severity of the defect is predicted, including setting the relationship model as a linear regression model, and calculating the output model y of the model, which is expressed as:

[0028] y=w 1 F depth,1 +w 2 F depth,2 +w 3 F depth,3 +b

[0029] Among them, w 1 , w 2 , w 3 is the weight parameter, b is the bias term, and w 1 F depth,1 +w 2 F depth,2 +w 3 F depth,3 is the deep waveform feature vector, y is the output of the model, and F depth,1 、F depth,2 and Fdepth,3 They are the components of the deep waveform feature vector.

[0030] As a preferred embodiment of the multi-modal mechanical wave detection method based on the boundary bounce characteristic of the present invention, wherein: the generation of the early warning scheme includes inputting the newly collected deep waveform features into the trained random forest model to obtain the prediction result, setting the early warning thresholds L1 and L2 according to the prediction result, and generating the early warning information. If y < L1, generate the early warning information for minor defects and recommend regular inspection and maintenance. If L1 ≤ y ≤ L2, generate the early warning information for medium defects and recommend immediate detailed inspection and taking necessary repair measures. If y ≥ L2, generate the early warning information for serious defects and recommend immediate stoppage of use and emergency repair or replacement.

[0031] Another object of the present invention is to provide a multi-modal mechanical wave detection system based on the boundary bounce characteristic, which can analyze the reflection characteristics when various modal mechanical waves encounter boundaries during propagation inside the target object, calculate the boundary bounce values of various modal mechanical waves, and select the mechanical wave with the highest boundary bounce value as the deep waveform feature, solving the problems of insufficient in-depth analysis of reflection characteristics and inaccurate calculation method of boundary bounce values in the current non-destructive testing technology.

[0032] As a preferred embodiment of the multi-modal mechanical wave detection system based on the boundary bounce characteristic of the present invention, wherein: it includes a data preprocessing module, a boundary bounce analysis module, and a defect prediction and early warning module.

[0033] The data preprocessing module is used to collect the waveforms, frequencies, and amplitudes of various modal mechanical waves by using sensors and perform preprocessing; the boundary bounce analysis module is used to analyze the reflection characteristics when various modal mechanical waves encounter boundaries during propagation inside the target object, calculate the boundary bounce values of various modal mechanical waves, and select the mechanical wave with the highest boundary bounce value as the deep waveform feature; the defect prediction and early warning module is used to construct a relationship model, input the deep waveform features into the relationship model, predict the severity of the defect, and generate the early warning scheme.

[0034] A computer device includes a memory and a processor, the memory stores a computer program, and the processor executes the computer program to implement the steps of the multi-modal mechanical wave detection method based on the boundary bounce characteristic.

[0035] A computer-readable storage medium stores a computer program thereon, and when the computer program is executed by a processor, it implements the steps of the multi-modal mechanical wave detection method based on the boundary bounce characteristic.

[0036] Beneficial effects of the present invention: The multimodal mechanical wave detection method based on boundary rebound characteristics provided by the present invention adopts sensors to collect the waveforms, frequencies and amplitudes of various modal mechanical waves and preprocess them, which effectively removes noise interference and improves signal clarity, laying a foundation for accurately analyzing the boundary rebound characteristics of mechanical waves, analyzing the reflection characteristics of various modal mechanical waves when they propagate inside the target object and encountering boundaries, calculating the boundary rebound values ​​of various modal mechanical waves, and selecting the mechanical wave with the highest boundary rebound value as the deep waveform feature, which can effectively identify which modal mechanical wave is most sensitive to boundary rebound, thereby selecting the most representative deep waveform feature, effectively removing noise and retaining the true reflection characteristics, and realizing the detection of mechanical waves in materials. The in-depth understanding of the internal propagation behavior of the material provides strong data support for the subsequent defect assessment, builds a relational model, inputs the deep waveform features into the relational model, predicts the severity of the defect, and generates an early warning plan, which converts complex waveform data into easy-to-understand defect severity indicators, providing a scientific basis for maintenance decisions. By setting early warning thresholds and generating corresponding early warning information, potential defects can be discovered in advance so that maintenance measures can be taken in time, and quantitative evaluation and real-time monitoring of internal corrosion defects of the target object are achieved, which improves the practicality and engineering application value of the detection system. The present invention achieves better results in terms of data acquisition accuracy, defect detection accuracy and the practicality of the early warning system. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0038] Figure 1 An overall flow chart of a multi-modal mechanical wave detection method based on boundary rebound characteristics provided for the first embodiment of the present invention.

[0039] Figure 2 An overall flow chart of a multi-modal mechanical wave detection system based on boundary rebound characteristics provided for the third embodiment of the present invention. DETAILED DESCRIPTION

[0040] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0041] Example 1, reference Figure 1 , as an embodiment of the present invention, provides a multi-modal mechanical wave detection method based on boundary rebound characteristics, comprising:

[0042] S1: Use sensors to collect the waveforms, frequencies and amplitudes of multiple modal mechanical waves and pre-process them.

[0043] Furthermore, sensors are used to collect waveforms, frequencies and amplitudes of multiple modal mechanical waves, including selecting MEMS sensors, configuring the sampling frequency and sampling time of the sensors, setting the time window for data acquisition, and collecting signal data of longitudinal waves, transverse waves, surface waves and plate waves when they propagate inside the target object.

[0044] It should be noted that the preprocessing process includes using the Butterworth filter to eliminate the noise interference of multiple modal mechanical waves and obtain the denoised signal x filtered (t), normalize the signal so that all signal values ​​fall within the same range, and calculate the normalized value x norm (t), expressed as:

[0045]

[0046] Among them, x min is the minimum value of the denoised signal, x max is the maximum value of the denoised signal, x norm (t) is the normalized value.

[0047] It should also be noted that by using high-precision sensors to collect the waveforms, frequencies and amplitudes of various modal mechanical waves and implementing effective preprocessing, the integrity and quality of the original data are ensured. The selection of MEMS sensors and configured sampling parameters can adjust the sampling frequency and time window according to different detection requirements, thereby adapting to various complex detection environments and improving the adaptability of detection. The filtering technology and normalization processing used in the preprocessing process effectively remove noise and interference in the signal, so that subsequent analysis can be based on purer signals, improving the accuracy of data analysis. The normalized signal obtained through preprocessing provides a unified benchmark for comparative analysis between different modal mechanical waves, helps to reveal the differences in the propagation of each modal wave within the material, and lays a solid foundation for subsequent reflection feature analysis and boundary rebound value calculation, ensuring the reliability and efficiency of the entire detection process.

[0048] S2: Analyze the reflection characteristics of multiple modal mechanical waves when they encounter boundaries when propagating inside the target object, calculate the boundary rebound values ​​of multiple modal mechanical waves, and select the mechanical wave with the highest boundary rebound value as the deep waveform feature.

[0049] Furthermore, the reflection behavior of multi-modal mechanical waves when encountering boundaries while propagating inside the target object is analyzed, including applying multi-modal mechanical waves at different positions of the target object, and using a high-speed camera and a laser Doppler vibrometer to record the response intensity of the waveform, and using the finite element analysis software COMSOL Multiphysics to simulate the response intensity of multi-modal waves at corrosion defects, and in the absence of corrosion defects, collecting the background noise signal q, and calculating the standard deviation σ of the background noise as an indicator of the noise level.

[0050] The reflection behavior of the waveform at different boundaries is recorded, and a multi-scale analysis method is introduced to analyze the boundary rebound characteristics at different scales from macro to micro, including using wavelet transform to remove noise and retain the main features of the waveform in the real reflection characteristics. The reflected wave field intensity S(t,S) after denoising is calculated and expressed as:

[0051]

[0052] Among them, s(t,s) is the reflected wave intensity after denoising, ψ is the wavelet basis function, s is the scale parameter, t is the time parameter, and x norm (τ) is the original signal, is an integral operation, dτ is an infinitesimal increment of the time variable τ, the signal data of the waveform characteristics are normalized to the same order of magnitude, and the normalized signal S is calculated norm (t,s), expressed as:

[0053]

[0054] Among them, S min is the minimum value of the signal after wavelet transformation, S max is the maximum value of the signal after wavelet transformation, S norm (t,s) is the normalized signal.

[0055] Collect the background noise signal Q and calculate the standard deviation σ of the background noise as an indicator of the noise level, including calculating the mean μ and the variance σ 2 , calculate the standard deviation σ, and evaluate the level of background noise based on the calculated σ. When σ exceeds the preset threshold σ1, it means that the background noise level is too high. Re-collect the background noise signal and use the adaptive filter to further denoise it. Calculate the denoised background noise signal W(t), which is expressed as:

[0056] W(t)=QH adapt (J)·E(t)

[0057] Where W(t) is the background noise signal after denoising, H adapt(J) is the transfer function of the adaptive filter, E(t) is the estimated noise signal, and the standard deviation σ2 of the denoised background noise signal is recalculated as:

[0058]

[0059] Among them, σ 2 is the standard deviation of the background noise signal after denoising, μ clean is the average value of the background noise signal after denoising, (W(t i )-μ clean ) 2 is the square of the difference between each denoised background noise signal value and the average value, is the summation symbol, W(t i ) is the value of the denoised background noise signal at the i-th time point, For the averaging operation, N is the length of the signal and i is each time point.

[0060] It should be noted that selecting the mechanical wave with the highest boundary rebound value as the depth waveform feature includes setting the boundary rebound value to R(t,s), which is expressed as:

[0061]

[0062] Among them, R(t,s) is the boundary rebound value, S norm (t, s) is the normalized reflected wave field strength. The mechanical wave with the highest boundary rebound value is selected as the depth waveform feature. The depth waveform feature F is calculated. depth , expressed as:

[0063]

[0064] Among them, F depth is the deep waveform feature, It means finding the values ​​of t and s that maximize R(t,s) over all times and scales.

[0065] It should also be noted that by applying multimodal mechanical waves at different positions of the target object and combining high-speed cameras and laser Doppler vibrometers to record waveform responses, the reflection behavior of mechanical waves is fully captured, solving the problem of insufficient in-depth analysis of reflection characteristics in the prior art. Finite element analysis software and multi-scale analysis methods are used to analyze boundary rebound characteristics at different scales from macro to micro, effectively identify small or complex defects, and improve the sensitivity and resolution of defect detection. By calculating and comparing the boundary rebound values ​​of mechanical waves of different modes, the most representative depth waveform features are selected, thereby providing a direct basis for the depth assessment of defects. By introducing adaptive filters for further denoising, the accuracy of signal processing is improved, ensuring the accuracy of boundary rebound value calculation.

[0066] S3: Construct a relationship model, input the depth waveform features into the relationship model, predict the severity of the defect, and generate a warning plan.

[0067] Furthermore, construct a relationship model based on the corrosion defect and the depth waveform features. Input the depth waveform features into the relationship model to predict the severity of the defect. Suppose the relationship model is a linear regression model, and the output model y of the model is calculated and expressed as:

[0068] y = w 1 F depth,1 + w 2 F depth,2 + w 3 F depth,3 + b

[0069] Among them, w 1 , w 2 , w 3 are weight parameters, b is the bias term, w 1 F depth,1 + w 2 F depth,2 + w 3 F depth,3 is the depth waveform feature vector, y is the output of the model, and F depth,1 , F depth,2 and F depth,3 are the respective components of the depth waveform feature vector.

[0070] It should be noted that generating a warning plan includes inputting the newly collected depth waveform features into the trained random forest model to obtain the prediction result, setting the warning thresholds L1 and L2 according to the prediction result, and generating a warning message. If y < L1, a warning message for minor defects is generated, and regular inspection and maintenance are recommended. If L1 ≤ y ≤ L2, a warning message for medium defects is generated, and detailed inspection is recommended immediately, and necessary repair measures are taken. If y ≥ L2, a warning message for severe defects is generated, and it is recommended to stop using immediately and perform emergency repair or replacement.

[0071] It should also be noted that by constructing a linear regression model or other suitable models, complex waveform data can be converted into an intuitive defect severity assessment, making the detection results easier to understand and apply. The relationship between deep waveform characteristics and corrosion defects is modeled, providing a method for quantitatively analyzing the severity of defects. Compared with traditional qualitative analysis, the prediction results are more reliable and specific. By setting early warning thresholds and generating early warning information, early detection and real-time monitoring of potential defects are achieved, providing timely and effective information support for maintenance decisions. The generation of early warning plans not only improves the practicality of the detection system, but also optimizes the prevention capabilities in actual engineering applications, which helps to reduce losses and risks caused by failure to detect defects in a timely manner.

[0072] Example 2 is an embodiment of the present invention, which provides a multi-modal mechanical wave detection method based on boundary rebound characteristics. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.

[0073] Firstly, six different types of test objects were selected, each with representative material properties. The test equipment included MEMS sensors, high-speed cameras, laser Doppler vibrometers, and finite element analysis software COMSOL Multiphysics. During the data acquisition phase, the sensors were configured to operate at a frequency of 1000 Hz and a sampling time of 5 seconds to collect signal data of longitudinal waves, transverse waves, surface waves, and plate waves propagating inside the target object. To ensure data quality, a Butterworth filter was used to preprocess the collected signals to eliminate noise interference, and normalization was performed to ensure that the signal values ​​fell within the same range. In terms of boundary rebound value calculation, the sensor was configured to operate at a frequency of 1000 Hz and a sampling time of 5 seconds to collect signal data of longitudinal waves, transverse waves, surface waves, and plate waves propagating inside the target object. In order to ensure data quality, a Butterworth filter was used to preprocess the collected signals to eliminate noise interference, and normalization was performed to ensure that the signal values ​​fell within the same range. , the denoised reflected wave field intensity is used as the basis for calculating the boundary rebound value, the mechanical wave with the highest boundary rebound value is selected as the deep waveform feature, and a calculation method based on the normalized reflected wave field intensity is provided, so that the deep waveform feature can more accurately reflect the defect information; finally, by constructing a relationship model and inputting the deep waveform feature into the model, the severity of the defect is predicted, and an early warning scheme is generated, which not only considers the relationship between the deep waveform feature and the severity of the defect, but also introduces a random forest model to improve the accuracy and reliability of the prediction; it can be seen from the experimental results that the present invention has advantages in improving signal quality, accurately calculating boundary rebound values, and generating early warning schemes.

[0074] Example 3, reference Figure 2 , which is an embodiment of the present invention, provides a multi-modal mechanical wave detection system based on boundary rebound characteristics, including a data preprocessing module, a boundary rebound analysis module, and a defect prediction and warning module.

[0075] The data preprocessing module is used to use sensors to collect the waveforms, frequencies and amplitudes of various modal mechanical waves and preprocess them; the boundary rebound analysis module is used to analyze the reflection characteristics of various modal mechanical waves when they encounter boundaries when propagating inside the target object, calculate the boundary rebound values ​​of various modal mechanical waves, and select the mechanical wave with the highest boundary rebound value as the deep waveform feature; the defect prediction and early warning module is used to construct a relationship model, input the deep waveform features into the relationship model, predict the severity of the defect, and generate an early warning plan.

[0076] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods of each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0077] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0078] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering, or processing in another suitable manner as necessary, and then stored in a computer memory.

[0079] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc. It should be noted that the above embodiments are only used to illustrate the technical solution of the present invention and are not limited. Although the present invention is described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention, which should be included in the scope of the claims of the present invention.

[0080] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A multi-modal mechanical wave detection method based on boundary rebound characteristics, characterized in that: Including: Using sensors to collect the waveforms, frequencies, and amplitudes of multiple modal mechanical waves and perform preprocessing; Analyzing the reflection characteristics of multiple modal mechanical waves when encountering boundaries during propagation inside the target object, calculating the boundary bounce values of multiple modal mechanical waves, and selecting the mechanical wave with the highest boundary bounce value as the depth waveform feature; Constructing a relationship model, inputting the depth waveform feature into the relationship model, predicting the severity of the defect, and generating a warning plan.

2. The multi-modal mechanical wave detection method based on boundary rebound characteristics according to claim 1, characterized in that: The step of using sensors to collect the waveforms, frequencies, and amplitudes of multiple modal mechanical waves includes selecting MEMS sensors, configuring the sampling frequency and sampling time of the sensors, setting the time window for data acquisition, and collecting the signal data of longitudinal waves, transverse waves, surface waves, and plate waves during propagation inside the target object.

3. The multi-modal mechanical wave detection method based on boundary rebound characteristics as claimed in claim 2, characterized in that: The preprocessing process includes using the filtering technology Butterworth filter to eliminate the noise interference of multiple modal mechanical waves and obtain the denoised signal x filtered (t), normalize the signal so that all signal values ​​fall within the same range, and calculate the normalized value x norm (t), expressed as: Among them, x min is the minimum value of the denoised signal, x max is the maximum value of the denoised signal, x norm (t) is the normalized value.

4. The multi-modal mechanical wave detection method based on boundary rebound characteristics as claimed in claim 3, characterized in that: The step of analyzing the reflection behavior of multiple modal mechanical waves when encountering boundaries during propagation inside the target object includes applying multi-modal mechanical waves at different positions of the target object, using a high-speed camera and a laser Doppler vibrometer to record the response intensity of the waveform, using the finite element analysis software COMSOL Multiphysics to simulate the response intensity of multiple modal waves at the corrosion defect, and collecting the background noise signal Q in the absence of corrosion defects, and calculating the standard deviation σ of the background noise as an index of the noise level; Recording the reflection behavior of the waveform at different boundaries, introducing a multi-scale analysis method, and analyzing the boundary bounce characteristics from different scales from macro to micro, including using wavelet transform to remove noise and retaining the main features in the true reflection characteristics of the waveform, and calculating the denoised reflected wave field strength S(t, s), expressed as: Among them, S(t,s) is the reflected wave intensity after denoising, ψ is the wavelet basis function, s is the scale parameter, t is the time parameter, x norm (τ) is the original signal, is an integral operation, dτ is an infinitesimal increment of the time variable τ, the signal data of the waveform characteristics are normalized to the same order of magnitude, and the normalized signal S is calculated norm (t,s), expressed as: Among them, S min is the minimum value of the signal after wavelet transformation, S max is the maximum value of the signal after wavelet transformation, S norm (t,s) is the normalized signal; Collect the background noise signal Q and calculate the standard deviation σ of the background noise as an indicator of the noise level, including calculating the mean μ and the variance σ 2 , calculate the standard deviation σ, and evaluate the level of background noise based on the calculated σ. When σ exceeds the preset threshold σ1, it means that the background noise level is too high. Re-collect the background noise signal and use the adaptive filter to further denoise it. Calculate the denoised background noise signal W(t), which is expressed as: W(t)=Q-H adapt (J)·E(t) Where W(t) is the background noise signal after denoising, H adapt (J) is the transfer function of the adaptive filter, E(t) is the estimated noise signal, and the standard deviation σ2 of the denoised background noise signal is recalculated as: Among them, σ 2 is the standard deviation of the background noise signal after denoising, μ clean is the average value of the background noise signal after denoising, (W(t i )-μ clean ) 2 is the square of the difference between each denoised background noise signal value and the average value, is the summation symbol, W(t i ) is the value of the denoised background noise signal at the i-th time point, For the averaging operation, N is the length of the signal and i is each time point.

5. The multi-modal mechanical wave detection method based on boundary rebound characteristics as claimed in claim 4, characterized in that: The step of selecting the mechanical wave with the highest boundary bounce value as the depth waveform feature includes setting the boundary bounce value as R(t, s), expressed as: Among them, R(t,s) is the boundary rebound value, S norm (t, s) is the normalized reflected wave field strength. The mechanical wave with the highest boundary rebound value is selected as the depth waveform feature. The depth waveform feature F is calculated. depth , expressed as: Among them, F depth is the deep waveform feature, It means finding the values ​​of t and s that maximize R(t,s) over all times and scales.

6. The multi-modal mechanical wave detection method based on boundary rebound characteristics as claimed in claim 5, characterized in that: The step of constructing a relationship model based on the corrosion defect and the depth waveform feature, inputting the depth waveform feature into the relationship model, and predicting the severity of the defect includes setting the relationship model as a linear regression model and calculating the output model y of the model, expressed as: y=w1F depth,1 +w2F depth,2 +w3F depth,3 +b Among them, w1, w2, w3 are weight parameters, b is the bias term, w1F depth,1 +w2F depth,2 +w3F depth,3 is the deep waveform feature vector, y is the output of the model, and F depth,1 、F depth,2 and F depth,3 are the various components of the depth waveform feature vector.

7. The multi-modal mechanical wave detection method based on boundary rebound characteristics according to claim 6, characterized in that: The step of generating a warning plan includes inputting the newly collected depth waveform feature into the trained random forest model to obtain the prediction result, setting the warning thresholds L1 and L2 according to the prediction result, and generating a warning message. If y < L1, a warning message for a minor defect is generated, suggesting regular inspection and maintenance. If L1 ≤ y ≤ L2, a warning message for a medium defect is generated, suggesting immediate detailed inspection and taking necessary repair measures. If y ≥ L2, a warning message for a severe defect is generated, suggesting immediate stoppage of use and emergency repair or replacement.

8. A system using the multi-modal mechanical wave detection method based on boundary rebound characteristics as claimed in any one of claims 1 to 7, characterized in that: Including a data preprocessing module, a boundary bounce analysis module, and a defect prediction and warning module; The data preprocessing module is used to collect the waveforms, frequencies, and amplitudes of multiple modal mechanical waves using sensors and perform preprocessing; The boundary bounce analysis module is used to analyze the reflection characteristics of multiple modal mechanical waves when encountering boundaries during propagation inside the target object, calculate the boundary bounce values of multiple modal mechanical waves, and select the mechanical wave with the highest boundary bounce value as the depth waveform feature; The defect prediction and early warning module is used to construct a relationship model, input the deep waveform features into the relationship model, predict the severity of the defect, and generate an early warning plan.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the multi-modal mechanical wave detection method based on boundary rebound characteristics described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the multi-modal mechanical wave detection method based on boundary rebound characteristics described in any one of claims 1 to 7 are implemented.

Images