Signal excitation and recognition method, system and computer device

Abstract

This application proposes a signal excitation and recognition method, system, and computer device. The method includes: constructing a model; performing modal analysis on the model to obtain natural frequencies and modes; determining the excitation period based on the natural frequencies of the structure under test; generating a periodic pulse excitation signal by a vibrator; generating vibration by impacting the test piece with a push rod; sensing the vibration by a sensor to form a raw vibration signal; then decomposing each raw vibration signal to obtain a combined signal based on the decomposed IMF components; extracting features from the combined signal to obtain corresponding feature parameter data; and constructing a dataset to train an initial signal recognition model, thereby completing the recognition of the test signal. The signal excitation and recognition method proposed in this application can monitor the structural health status in real time and accurately, thereby detecting potential structural faults in advance and improving the safety and efficiency of equipment operation.

Description

Technical Field

[0001] This application relates to the field of structural health monitoring technology, and in particular to a signal excitation and recognition method, system and computer equipment. Background Technology

[0002] Structural health monitoring and diagnostics are extremely important in aerospace, medical, and structural engineering fields. It typically requires high-precision sensing technologies to detect even minute changes; for example, in the medical field, loosening of artificial bone structures can severely impact treatment outcomes.

[0003] Vibration devices used to monitor structural loosening generally employ active excitation, which can be categorized into hammer excitation and vibrator excitation based on the excitation tool. Hammer excitation applies a very short-duration excitation signal to the test object only once; from a signal perspective, this excitation signal approximates a high-energy impulse function. Because the excitation energy is supplied only once, it requires a large amount of energy and may damage the test object. For small-scale, low-mass structural components, this excitation method is unsuitable for long-term structural health monitoring.

[0004] In excitation methods using a vibrator, the output force rod of the vibrator is typically connected to the test object (DAMP) via a connector. In this setup, the excitation rod is connected to the DAMP through the connector. The excitation signal can be deterministic or random. Deterministic signals generally include sinusoidal sweep excitation and sinusoidal fast sweep excitation. Random excitation signals include pure random excitation, windowed random excitation, pseudo-random excitation, periodic random excitation, and burst random excitation. The excitation of the structure under test by these signals is based on the physical structure of the vibrator setup. However, for small-mass, small-scale structural components, the mass or size of the connecting parts is on the same order of magnitude or larger than that of the DAMP. Therefore, this typical vibrator setup directly affects the dynamic characteristics of the DAMP, making it unsuitable for monitoring small-mass, small-scale structural components. Summary of the Invention

[0005] Based on this, the purpose of this application is to propose a signal excitation and recognition method, system and computer equipment, which is based on the use of a suspended FBG sensing system, in order to solve the problem that traditional monitoring methods are difficult to apply to small-mass, small-scale structural components.

[0006] This application proposes a signal excitation and recognition method, the method comprising:

[0007] A finite element simulation analysis model of the structure under test is constructed, and modal analysis is performed on the finite element simulation analysis model of the structure under test to obtain the natural frequency and mode of the structure under test. Based on this, the assembly method between the suspended fiber Bragg grating sensor and the structure under test is determined, and the resonant frequency of the suspended fiber Bragg grating sensor is obtained according to the assembly method between the suspended fiber Bragg grating sensor and the structure under test.

[0008] The excitation period is determined according to the natural frequency. The exciter generates a periodic pulse excitation signal according to the excitation period. The original vibration signal is generated by impacting the test structure with a known state signal through the push rod according to the periodic pulse excitation signal. The known state signal includes at least torque signal, crack signal, friction coefficient signal, and material purity signal.

[0009] The original vibration signal is decomposed and extracted using the empirical mode decomposition algorithm to obtain multi-order IMF components, and the cross-correlation coefficient between the time-domain signal of each IMF component and the original vibration signal is calculated.

[0010] At least one target IMF component is selected from all IMF components based on the cross-correlation coefficient, and the target IMF component with suspended fiber Bragg grating resonance noise is denoised based on the resonant frequency of the suspended fiber Bragg grating sensor. After denoising, all target IMF components are superimposed to obtain a combined signal.

[0011] Dimensional and dimensionless parameters are extracted from the combined signal, and the dimensional and dimensionless parameters are filtered to obtain feature parameter data, so as to construct a dataset based on the feature parameter data.

[0012] The dataset is input into the initial signal recognition model for training to obtain the final signal recognition model. The test signal is then input into the final signal recognition model to obtain the health status classification result.

[0013] In some embodiments, the steps of constructing a finite element simulation analysis model of the structure under test, performing modal analysis on the finite element simulation analysis model of the structure under test to obtain the natural frequencies and modes of the structure under test, thereby determining the assembly method between the suspended fiber Bragg grating sensor and the structure under test, and obtaining the resonant frequency of the suspended fiber Bragg grating sensor based on the assembly method between the suspended fiber Bragg grating sensor and the structure under test include:

[0014] The adhesive point position and the fiber Bragg grating position are determined based on the modal analysis results, and the adhesive length, the distance from the adhesive point to the fiber Bragg grating, and the distance from the fiber Bragg grating to the pigtail are obtained based on the adhesive point position and the fiber Bragg grating position.

[0015] The total length L is calculated based on the distance from the adhesive point to the fiber Bragg grating, the distance from the fiber Bragg grating to the pigtail, and the grating length of the fiber Bragg grating. The resonant frequency of the suspended fiber Bragg grating sensor is then calculated based on the total length L.

[0016] In some embodiments, the steps of determining the excitation period based on the natural frequency, generating a periodic pulse excitation signal by a vibrator based on the excitation period, and generating an original vibration signal by striking a structural member under test with a known state signal using a push rod based on the periodic pulse excitation signal include:

[0017] The excitation signal is obtained according to the following formula:

[0018]

[0019] in, Indicates the excitation signal. Represents the unit impulse function. N is an integer, A represents the impulse intensity, t represents time, and T represents the excitation period;

[0020] The step of calculating the resonant frequency of the suspended fiber Bragg grating sensor based on the total length L includes:

[0021] The resonant frequency of the suspended fiber Bragg grating sensor is calculated using the following formula:

[0022]

[0023] Where f0 represents the resonant frequency of the suspended fiber Bragg grating sensor, and C represents the velocity of sound waves in the fiber.

[0024] In some embodiments, the step of using an empirical mode decomposition algorithm to decompose and extract the original vibration signal to obtain multi-order IMF components includes:

[0025] The original vibration signal is decomposed into two parts: local smoothing and local oscillation. The extreme points of the local smoothing part are extracted to obtain a set of local extreme point sequences. The local extreme point sequences are then interpolated and fitted to obtain a set of local smoothing functions.

[0026] A set of local oscillation functions is obtained based on the local smoothing function and the original vibration signal, and the first-order IMF component is obtained based on the local oscillation function;

[0027] The remaining components after the first decomposition are obtained based on the first-order IMF components and the original vibration signal. The remaining components after the first decomposition are then repeatedly decomposed and extreme points are extracted to obtain the second-order IMF components.

[0028] Repeat the above steps until the k-th order IMF component is obtained.

[0029] In some embodiments, the step of obtaining a set of local oscillation functions based on the local smoothing function and the original vibration signal, and obtaining the first-order IMF component based on the local oscillation functions, includes:

[0030] The local oscillation function can be obtained using the following formula:

[0031]

[0032] in, Represents a local oscillation function. Represents the original vibration signal. Represents a local smoothing function;

[0033] Determine whether the local oscillation function satisfies the preset IMF component conditions, which include the number of extreme points in the local oscillation function being equal to the number of zero crossover points and the average value of the envelope function being zero;

[0034] If the local oscillation function does not meet the preset IMF component conditions, the obtained local oscillation function is used as a new vibration signal, and a new local oscillation function is calculated again based on the new vibration signal until the new local oscillation function meets the preset IMF component conditions. Then, the local oscillation function that meets the preset IMF component conditions is output as the first-order IMF component.

[0035] If the local oscillation function satisfies the preset IMF component conditions, then the local oscillation function will be... As a first-order IMF component;

[0036] The k-th order IMF component is obtained using the following formula:

[0037]

[0038] Where n represents the number of decompositions. Represents the k-th order IMF component. This represents the remaining components after the nth decomposition.

[0039] In some embodiments, the step of calculating the cross-correlation coefficient between the time-domain signal of each IMF component and the original vibration signal includes:

[0040] The cross-correlation coefficient can be obtained using the following formula:

[0041]

[0042] The mean of the time-domain signal of all sampled k-th order IMF components is obtained using the following formula:

[0043]

[0044] The mean of the time-domain signal of all sampled original vibration signals is obtained using the following formula:

[0045]

[0046] Where r represents the cross-correlation coefficient between the time-domain signal of the k-th IMF component and the time-domain signal of the original vibration signal, i represents the i-th sampling point, and m represents the total number of sampling points. This represents the time-domain signal of the i-th sampling point of the k-th IMF component. Let represent the mean of the time-domain signal of all sampled k-th order IMF components. This represents the mean of the time-domain signal of all sampled original vibration signals. This represents the time-domain signal of the original vibration signal at the i-th sampling point.

[0047] In some embodiments, the steps of selecting at least one target IMF component from all IMF components based on the cross-correlation coefficient, denoising the target IMF component with suspended fiber Bragg grating resonance noise based on the resonant frequency of the suspended fiber Bragg grating sensor, and then superimposing all the target IMF components to obtain a combined signal include:

[0048] Determine whether the cross-correlation coefficient corresponding to any IMF component of any order is greater than a first preset threshold, so that all IMF components that are greater than the first preset threshold are taken as target IMF components.

[0049] The filtering frequency range is determined based on the resonant frequency of the suspended fiber Bragg grating sensor, so as to denoise the target IMF component with suspended fiber Bragg grating resonant noise according to the filtering frequency range.

[0050] In some embodiments, the step of extracting dimensional parameters and dimensionless parameters from the combined signal, and filtering the dimensional parameters and the dimensionless parameters to obtain feature parameter data includes:

[0051] The dimensional parameters include mean, standard deviation, maximum, minimum, residual, peak-to-peak value, and energy; the dimensionless parameters include skewness, kurtosis, waveform factor, amplitude factor, impact factor, and margin factor.

[0052] Principal component analysis (PCA) was used to reduce the dimensionality of all dimensional and dimensionless parameters to eliminate redundant variables and obtain characteristic parameter data.

[0053] This application also provides a signal excitation and recognition system, the system comprising:

[0054] The finite element modal analysis module is used to build a finite element simulation analysis model of the structure under test and perform modal analysis on the finite element simulation analysis model of the structure under test to obtain the natural frequency and mode of the structure under test, thereby determining the assembly method between the suspended fiber Bragg grating sensor and the structure under test, and obtaining the resonant frequency of the suspended fiber Bragg grating sensor according to the assembly method between the suspended fiber Bragg grating sensor and the structure under test.

[0055] The excitation signal generation module is used to determine the excitation period according to the natural frequency, generate a periodic pulse excitation signal by the exciter according to the excitation period, and generate an original vibration signal by impacting the test structure with a known state signal through the push rod according to the periodic pulse excitation signal. The known state signal includes at least torque signal, crack signal, friction coefficient signal, and material purity signal.

[0056] The vibration signal decomposition module is used to decompose and extract the original vibration signal using the empirical mode decomposition algorithm to obtain multi-order IMF components, and to calculate the cross-correlation coefficient between the time domain signal of each IMF component and the original vibration signal.

[0057] The signal denoising module is used to filter at least one target IMF component from all IMF components according to the cross-correlation coefficient, and to denoise the target IMF component with suspended fiber Bragg grating resonance noise according to the resonance frequency of the suspended fiber Bragg grating sensor. After denoising, all target IMF components are superimposed to obtain a combined signal.

[0058] The dataset construction module is used to extract dimensional parameters and dimensionless parameters from the combined signal, and to filter the dimensional parameters and the dimensionless parameters to obtain feature parameter data, so as to construct a dataset based on the feature parameter data.

[0059] The recognition result output module is used to input the dataset into the initial signal recognition model for training to obtain the final signal recognition model, and input the signal to be tested into the final signal recognition model to obtain the health status classification result.

[0060] This application also provides a storage medium that stores one or more programs that, when executed, implement the signal excitation and recognition method as described above.

[0061] This application also provides a computer device, the computer device including a memory and a processor, wherein:

[0062] The memory is used to store computer programs;

[0063] When the processor executes the computer program stored in the memory, it implements the signal excitation and recognition method described above.

[0064] Compared with the prior art, this application has the following advantages:

[0065] (1) The physical realization of the excitation signal in this application does not require a connecting device; that is, the exciter rod can be used to give the structure under test an impulse excitation. There is no need for a connecting device between the exciter rod and the test piece to generate actual physical excitation. This setting only requires the rod to have a very brief contact with the test piece under the action of the signal generator to generate an impulse excitation signal.

[0066] (2) This application achieves selective coverage of the target frequency by periodically repeating the pulse excitation, and the energy of each frequency component is the same. Therefore, by reasonably selecting the excitation frequency, the excitation signal energy of each target frequency component can be consistent.

[0067] (3) Since the excitation coverage frequency of this application is discretely distributed rather than continuously distributed, the frequency components of the effective signal in signal analysis can be effectively reduced, the difficulty of signal analysis can be reduced, and thus the signal recognition accuracy can be improved.

[0068] (4) Since the excitation signal energy can be continuously supplied, from the perspective of energy conservation, the output energy can be controlled within a small range. This effectively avoids the problem that when the traditional monitoring method uses a hammer to perform a single action, a large impulse signal must be given in order to excite an effective signal, which may lead to structural damage.

[0069] (5) This application decomposes the vibration signal and performs targeted noise reduction on the IMF component obtained after decomposition, thereby avoiding the noise introduced by the dynamic characteristics of the sensing device itself and effectively improving the accuracy of signal recognition.

[0070] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by means of embodiments thereof. Attached Figure Description

[0071] Figure 1 This is a schematic diagram of the structure of the suspended FBG-based structural resonance-fiber acoustic waveguide sensing system in this application;

[0072] Figure 2This is a flowchart of the signal excitation and recognition method proposed in the first embodiment of this application;

[0073] Figure 3 This is a schematic diagram of the original vibration signal obtained by the suspended FBG sensing system in the first embodiment of this application;

[0074] Figure 4 This is a schematic diagram of the signals of each order of IMF components obtained after decomposition in the first embodiment of this application;

[0075] Figure 5 This is a spectrum diagram of the third-order IMF component in the first embodiment of this application;

[0076] Figure 6 This is the denoised spectrum of the third-order IMF component in the first embodiment of this application;

[0077] Figure 7 This is a schematic diagram of the excitation system used in the first embodiment of this application;

[0078] Figure 8 This is a schematic diagram of the excitation system in traditional technology;

[0079] Figure 9 This is a schematic diagram of the signal excitation and recognition system proposed in the second embodiment of this application;

[0080] Figure 10 This is a time-domain diagram of the excitation signal in the first embodiment of this application;

[0081] Figure 11 This is a frequency domain diagram of the excitation signal in the first embodiment of this application.

[0082] Symbol explanation: 1-exciter, 2-tunable narrowband laser source, 3-isolator, 4-fiber circulator, 5-photodetector, 6-data acquisition unit, 7-data analyzer, 8-second pigtail, 9-fiber Bragg grating, 10-first pigtail, 11-test structure.

[0083] The following detailed description, in conjunction with the accompanying drawings, will further illustrate this application. Detailed Implementation

[0084] To facilitate understanding of this application, a more complete description will be provided below with reference to the accompanying drawings, which illustrate several embodiments of the present application. However, the present application can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided to make the disclosure of this application more thorough and complete.

[0085] Unless otherwise defined, all technical and scientific terms used in this application have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used in this application and in its specification is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. The term "and / or" as used in this application includes any and all combinations of one or more of the associated listed items.

[0086] Structural health monitoring and diagnostics are extremely important in aerospace, medical, and structural engineering fields. It typically requires high-precision sensing technologies to detect even minute changes; for example, in the medical field, loosening of artificial bone structures can severely impact treatment outcomes.

[0087] Currently, mainstream structural health monitoring systems primarily utilize structural dynamics for monitoring. For overall structural monitoring, structural vibration sensing technology is mainly used, while acoustic / ultrasonic technology is primarily used for monitoring local structures. For sensing dynamic characteristics, piezoelectric devices and strain gauges are mainly used. Piezoelectric devices generate current by introducing pressure into a material, while strain gauges detect stress by measuring the deformation of the material. These sensors operate on essentially the same principle: obtaining information by measuring changes in physical quantities (such as pressure, displacement, and temperature). However, piezoelectric devices and strain gauges are electronic devices, which may be susceptible to external electromagnetic interference and cannot operate in harsh environments such as high temperature, high humidity, and high salinity. Furthermore, these devices have relatively large geometric dimensions; even small piezoelectric devices and strain gauges typically have sensing surface dimensions on the order of millimeters to tens of millimeters. Therefore, for small-scale, low-mass structural components, the mass of these sensors may be greater than the structure being measured. For example, in detecting the loosening of hollow screws in artificial bone blocks (which are only a few millimeters in size and weigh only a few grams), piezoelectric devices and strain gauges are not suitable as sensors.

[0088] Fiber optic sensors, with diameters on the order of micrometers and small mass, can significantly reduce the impact on the structure under test. Based on the above, this application uses suspended fiber Bragg grating (FBG) structural vibration-acoustic waveguide sensing technology to monitor the structure under test. Please refer to [link to relevant documentation]. Figure 1The diagram shows a structural resonance-fiber acoustic waveguide sensing system based on a suspended FBG. Specifically, a periodic pulse excitation signal generated by an exciter 1 via a push rod acts on the structure under test 11, generating a vibration signal. The first end of the FBG9, a pigtail 10, is attached to the structure under test 11 to sense the vibration. The vibration signal carrying structural health information at the attachment point forms a rod wave, which propagates on the second pigtail 8 and FBG9. The FBG senses this rod wave, and the FBG demodulator, composed of an adjustable narrowband laser source 2, an isolator 3, a fiber optic circulator 4, a photodetector 5, a data acquisition unit 6, and a data analyzer 7, outputs a test signal. The health status of the structure is assessed through signal analysis. The main purpose of using a suspended FBG sensor attachment method is to avoid distortion of the FBG's spectral performance caused by direct attachment. Furthermore, if the test component is in a harsh environment (e.g., high temperature), indirect attachment prevents performance degradation of the FBG due to the harsh environment. In summary, this sensing system is essentially a vibration information sensor for structural components, and is particularly suitable for extracting vibration information from small-scale, low-mass structural components.

[0089] In the field of structural health monitoring, signal excitation and signal processing directly affect monitoring quality. Excitation signals are used to stimulate the dynamic characteristics of the structure under test, while response signals are used to assess the health status of the structure and identify potential defects. The principle for selecting excitation signals is to elicit effective structural dynamic characteristics for subsequent structural health monitoring while minimizing the impact of the excitation on the structure. The purpose of selecting response signal identification methods is to accurately identify potential defects; the primary criterion is the accuracy of identification.

[0090] Please see Figure 2 The diagram shows a flowchart of a signal excitation and recognition method according to a first embodiment of this application. The method includes steps S01 to S06, wherein:

[0091] Step S01: Build a finite element simulation analysis model of the structure under test, and perform modal analysis on the finite element simulation analysis model of the structure under test to obtain the natural frequency and mode of the structure under test, thereby determining the assembly method between the suspended fiber Bragg grating sensor and the structure under test, and obtaining the resonant frequency of the suspended fiber Bragg grating sensor according to the assembly method between the suspended fiber Bragg grating sensor and the structure under test.

[0092] In this step, the assembly method includes the adhesive point position and the fiber Bragg grating position, and the adhesive length, the distance from the adhesive point to the fiber Bragg grating, the distance from the fiber Bragg grating to the pigtail, and the grating length of the fiber Bragg grating are obtained based on the adhesive point position and the fiber Bragg grating position.

[0093] Furthermore, the total length L is calculated by summing the distance from the adhesive point to the fiber Bragg grating, the distance from the fiber Bragg grating to the pigtail, and the grating length of the fiber Bragg grating.

[0094] Examples rather than limitations, combined Figure 1 As shown, the total length L refers to the distance from the end of the first pigtail furthest from the adhesive point on the test structure to the end of the second pigtail furthest from the fiber Bragg grating.

[0095] Furthermore, after obtaining the total length L, the resonant frequency of the suspended fiber Bragg grating sensor can be calculated using the following formula:

[0096]

[0097] in, The resonant frequency of the suspended fiber Bragg grating sensor is represented by , and C represents the velocity of sound waves in the fiber.

[0098] Step S02: Determine the excitation period according to the natural frequency, generate a periodic pulse excitation signal by the exciter according to the excitation period, and generate the original vibration signal by impacting the test structure with a known state signal through the push rod according to the periodic pulse excitation signal;

[0099] It should be noted that the known state signals include at least torque signals, crack signals, friction coefficient signals, and material purity signals. In other words, the solution adopted in this embodiment can be applied to the detection of torque signals, crack signals, friction coefficient signals, and material purity signals. For example, if the known state signal is a torque signal, structural components with known torque values ​​such as 0 N·cm, 10 N·cm, 20 N·cm, 30 N·cm, and 50 N·cm can be selected. If the known state signal is a crack signal, structural components with known crack width values ​​such as 1 mm, 2 mm, and 3 mm can be selected to impact the structural component with the known state signal to generate the original vibration signal.

[0100] It should be noted that the time-domain diagram of this excitation signal is as follows: Figure 10 As shown, the excitation period is determined based on the natural frequency, and then the excitation signal is obtained according to the following formula:

[0101]

[0102] in, Indicates the excitation signal. Represents the unit impulse function. N is an integer, t represents time, T represents the excitation period, and A represents the impulse intensity. The principle for selecting the value of A is to minimize the effective sensing signal obtained by the FBG, so as to protect the structure under test from damage due to the external force load during the test to the greatest extent.

[0103] Furthermore, after the excitation signal undergoes a Fourier transform in a linear system, its spectrum function can be obtained as follows:

[0104]

[0105] Its frequency domain graph corresponding to the spectral function is as follows Figure 11 As shown, where, The angular frequency of the excitation signal, Represents the unit impulse function. This represents the angular frequency, where m is an integer.

[0106] The advantages of using this technology are as follows:

[0107] First, the physical realization of this excitation signal does not require a connecting device; the exciter push rod can be used to apply impulse excitation to the structure under test, and no connecting device is needed between the exciter push rod and the test piece to generate actual physical excitation. This setup only requires the push rod to have a very brief contact with the test piece under the action of the signal generator to generate an excitation signal, effectively solving the problem that in the health monitoring of small-mass, small-size structural components, the mass and dimensions of the connecting device itself can be comparable to or larger than the mass and dimensions of the test piece.

[0108] Second, since the spectrum of the excitation signal is:

[0109]

[0110] This is the equal-amplitude superposition of the excitation frequency harmonics. Therefore, by reasonably selecting the excitation frequency, selective coverage of the target frequency can be achieved, and the excitation signal energy is consistent on each target frequency component.

[0111] Third, since the excitation coverage frequency is discretely distributed rather than continuously distributed, this can effectively reduce the frequency components of the effective signal in signal analysis and reduce the difficulty of signal analysis.

[0112] Fourth, since the energy supply for the excitation signal is continuous, from the perspective of energy conservation, the output energy can be relatively small. This effectively avoids the problem that when the hammer only performs a single action, a large impulse signal must be given to generate an effective signal, which may lead to structural damage.

[0113] Fifth, as mentioned in point four, a low single-impulse intensity of the excitation signal can effectively avoid nonlinear effects in structural dynamics caused by an excessively strong excitation signal. This can effectively reduce the difficulty of processing the response signal.

[0114] It should be noted that by constructing a finite element simulation analysis model of the structure under test, modal analysis is performed on the model to obtain the natural frequency and modes. The excitation period T is determined based on the natural frequency. The exciter generates a periodic pulse signal through the push rod to impact the structure under test, generating vibration on the structure under test. The suspended fiber Bragg grating sensor installed on the structure under test senses this vibration and forms the original vibration signal.

[0115] Please see Figure 7 The diagram shows a schematic of the excitation system used in this application. The component under test (DUT) refers to the structural component or the structural component to be tested. As can be seen from the diagram, no connecting device is used between the exciter push rod and the DUT. Please refer to [link to relevant documentation]. Figure 8 The diagram shows a schematic of the excitation system in a traditional technique. As can be seen, in current technology, the exciter push rod is connected to the test piece (DPT) via a connector. When the DPT is a small-mass, small-sized structure, the mass and size of the connector can affect the dynamic characteristics of the DPT itself, introducing significant noise signals into the test. Therefore, it should be noted that because the excitation coverage frequency is discretely distributed rather than continuously distributed, this effectively reduces the frequency components of the effective signal in signal analysis, thus reducing the difficulty of signal analysis. Simultaneously, because the excitation signal energy supply is continuous, from the perspective of energy conservation, a single output energy can excite effective vibration with a relatively small amount of energy. This excitation method effectively avoids the problem that when the hammer only performs a single action, a large impulse signal must be provided to excite an effective signal, which could potentially damage the structure. Furthermore, because the excitation signal generated in this embodiment has a low single impulse intensity, it effectively avoids the nonlinear effects of structural dynamics caused by an excessively strong excitation signal, thereby effectively reducing the difficulty of response signal processing.

[0116] As an example, and not a limitation, in some embodiments of this application, the excitation period is determined to be 314Hz. A total of 500 sets of vibration signals are collected using a suspended FBG (sampling time is 0.05 seconds, each signal has 3000 data points). Fasteners with different torques are embedded in the test piece, where the known torques are 0 N·cm, 10 N·cm, 20 N·cm, 30 N·cm, 50 N·cm, etc., with 100 sets for each known torque. Taking the original vibration signal of a bolt (structural component) under a torque of 30 N·cm as an example, such as... Figure 3 As shown.

[0117] Step S03: The original vibration signal is decomposed and extracted using the empirical mode decomposition algorithm to obtain multi-order IMF components, and the cross-correlation coefficient between the time domain signal of each IMF component and the original vibration signal is calculated.

[0118] In this step, the Empirical Mode Decomposition (EMD) algorithm is used to decompose the original signal and adaptively generate a set of Intrinsic Mode Functions (IMFs). The resulting IMF components of each order are as follows: Figure 4 As shown.

[0119] The main idea of ​​Empirical Mode Decomposition (EMD) is to decompose the vibration signal generated by a bolted structure into a signal composed of multiple IMF components. These IMF components represent different specific time scales of the signal. Since the IMF components derived from EMD are derived from the original vibration signal itself, the results guarantee the complete non-stationarity of the considered signal, providing a good analytical effect on the original signal.

[0120] In some embodiments, the specific decomposition process of the original vibration signal is as follows:

[0121] The original vibration signal is decomposed into two parts: local smoothing and local oscillation. The extreme points of the local smoothing part are extracted to obtain a set of local extreme point sequences. The local extreme point sequences are then interpolated and fitted to obtain a set of local smoothing functions.

[0122] A set of local oscillation functions is obtained based on the local smoothing function and the original vibration signal, and the first-order IMF component is obtained based on the local oscillation function;

[0123] The remaining components after the first decomposition are obtained based on the first-order IMF components and the original vibration signal. The remaining components after the first decomposition are then repeatedly decomposed and extreme points are extracted to obtain the second-order IMF components.

[0124] Repeat the above steps until the k-th order IMF component is obtained. The IMF components obtained after decomposition are as follows: Figure 4 As shown.

[0125] Specifically, the local oscillation function is obtained according to the following formula:

[0126]

[0127] in, Represents a local oscillation function. Represents the original vibration signal. Represents a local smoothing function;

[0128] Then it is determined whether the local oscillation function satisfies the preset IMF component conditions, which include the number of extreme points in the local oscillation function being equal to the number of zero crossover points and the average value of the envelope function being zero.

[0129] If the local oscillation function does not meet the preset IMF component conditions, the obtained local oscillation function is used as a new vibration signal, and a new local oscillation function is calculated again based on the new vibration signal until the new local oscillation function meets the preset IMF component conditions. Then, the local oscillation function that meets the preset IMF component conditions is output as the first-order IMF component.

[0130] If the local oscillation function satisfies the preset IMF component conditions, then the local oscillation function will be... As a first-order IMF component;

[0131] The k-th order IMF component is obtained using the following formula:

[0132]

[0133] Where n represents the number of decompositions. Represents the k-th order IMF component. This represents the remaining components after the nth decomposition.

[0134] It should also be noted that after decomposing each IMF component, it is necessary to calculate the cross-correlation coefficient between the time-domain signal of each IMF component and the original vibration signal.

[0135] In some embodiments, the cross-correlation coefficient is obtained according to the following formula:

[0136]

[0137] The mean of the time-domain signal of all sampled k-th order IMF components is obtained using the following formula:

[0138]

[0139] The mean of the time-domain signal of all sampled original vibration signals is obtained using the following formula:

[0140]

[0141] Where r represents the cross-correlation coefficient between the time-domain signal of the k-th IMF component and the time-domain signal of the original vibration signal, i represents the i-th sampling point, and m represents the total number of sampling points. This represents the time-domain signal of the k-th order IMF component at the i-th sampling point. Let represent the mean of the time-domain signal of all sampled k-th order IMF components. This represents the mean of the time-domain signal of all sampled original vibration signals. This represents the time-domain signal of the original vibration signal at the i-th sampling point.

[0142] As an example, not a limitation, the cross-correlation coefficients between each order of IMF components and the original vibration signal are shown in Table 1. The cross-correlation analysis of the second to fourth order IMF components and the original vibration signal shows that the variation trends of the second to fourth order IMF components are highly similar to the variation trends of the original vibration signal and have a high correlation with the original signal, indicating a close internal connection.

[0143] Table 1: Cross-correlation coefficients between IMF components of each order and the original vibration signal

[0144]

[0145] Step S04: Select at least one target IMF component from all IMF components according to the cross-correlation coefficient, and denoise the target IMF component with suspended fiber Bragg grating resonance noise according to the resonance frequency of the suspended fiber Bragg grating sensor. After denoising, superimpose all target IMF components to obtain a combined signal.

[0146] In some embodiments, after obtaining the cross-correlation coefficients corresponding to each IMF component, it is also necessary to determine whether the cross-correlation coefficients corresponding to any IMF component are greater than a first preset threshold, so that all IMF components that are greater than the first preset threshold are taken as target IMF components.

[0147] The filtering frequency range is determined based on the resonant frequency of the suspended fiber Bragg grating sensor, so as to denoise the target IMF component with resonant noise according to the filtering frequency range.

[0148] For example, not as a limitation, the first preset threshold can be set to 0.8. Based on the cross-correlation coefficients obtained from Table 1 above, the 2nd to 4th order IMF components are selected as target IMF components. In this embodiment, L is 0.8m, and C is the velocity of sound in the optical fiber, which is 3783.88m / s. Therefore, the resonant frequency of the suspended fiber Bragg grating sensor in this embodiment is... The frequency is 1182.5Hz. FBG amplifies signals near the resonant frequency of this suspended fiber Bragg grating sensor; therefore, in this embodiment, it is chosen as 1182.5Hz. As the frequency range where fiber resonance has a significant impact on the test signal, in this example, the filtered frequency range is 945-1418Hz.

[0149] Further, after determining the filter frequency range, please refer to [link / reference needed]. Figure 5 Based on this range, the spectrum of the third-order IMF component contains noise caused by the resonance of the suspended fiber Bragg grating sensor. To improve the recognition accuracy, this noise needs to be removed. In some embodiments of this application, this signal is specifically filtered out by introducing a Butterworth IIR band-stop filter. The spectrum of the third-order IMF component after filtering is as follows: Figure 6 As shown, the signals of the denoised 3rd-order IMF component, the 2nd-order IMF components, and the 4th-order IMF components are then directly superimposed to obtain the combined signal. In other words, the IMF component with the resonant frequency of the suspended fiber Bragg grating sensor is denoised to remove the sensor's influence on the signal. After denoising, all target IMF components are then superimposed to obtain the combined signal.

[0150] Step S05: Extract dimensional parameters and dimensionless parameters from the combined signal respectively, and filter the dimensional parameters and the dimensionless parameters to obtain feature parameter data, so as to construct a dataset based on the feature parameter data;

[0151] To more accurately analyze the various characteristic parameters of the combined signal, based on the stress conditions of the bolts (structural components) and the structure, and the characteristics of the collected vibration signals, this application selects a combination of dimensional and dimensionless characteristic parameters to jointly reflect and analyze the characteristics of the filtered 2nd-4th order IMF component superposition time-domain signal. The magnitude of the dimensional characteristic values ​​changes with the bolt connection state under different torques and is also affected by temperature, humidity, and other environmental factors, thus reflecting the bolt connection state under different torques to a certain extent. Meanwhile, dimensionless indices are insensitive to strain factors and changes in the working environment during vibration, and can more accurately reflect the characteristics of vibration signals under different bolt connection states. Therefore, this application combines dimensional and dimensionless characteristic indices for analysis.

[0152] As vibration time increases, not only does the contact area between the test structure and the vibrator change, but environmental factors such as temperature and humidity also fluctuate. Therefore, combining dimensional and dimensionless eigenvalues ​​as characteristic parameters is more adaptable and representative. Furthermore, changing the torque and thus altering the bolt connection state will significantly change the amplitude, energy, and power spectrum of the vibration signal, resulting in substantial changes in both dimensional and dimensionless eigenvalues.

[0153] In some embodiments, this application may select seven dimensional characteristic parameters (mean, standard deviation, maximum, minimum, residual, peak-to-peak value, and energy) and six dimensionless characteristic parameters (skewness, kurtosis, waveform factor, amplitude factor, impulse factor, and margin factor) as training sample feature components for time-domain analysis. The calculation results of various parameters for a set of signals with a torque of 30 N·cm are shown in Table 2 below.

[0154] Table 2. Thirteen dimensional and dimensionless characteristic parameters of the signal when the torque is 30 N·cm.

[0155]

[0156] Then, Principal Component Analysis (PCA) is used to reduce the dimensionality of all dimensional and dimensionless parameters to remove redundant variables and obtain the characteristic parameter data. The following is the PCA process:

[0157] 1. Standardize the feature matrix to obtain the standardized feature matrix.

[0158] 2. Calculate the covariance matrix of the standardized feature matrix.

[0159] 3. Calculate the eigenvalues ​​and eigenvectors of the covariance matrix to obtain the eigenvector matrix and eigenvalue matrix.

[0160] 4. Sort the eigenvalues ​​and obtain the sorted eigenvector matrix based on the sorted eigenvector matrix.

[0161] 5. Determine the number of primary features, calculate the cumulative variance contribution rate, and find the number of primary features that meet the threshold of 95%.

[0162] 6. Select the first 10 eigenvectors as the principal eigenvectors to obtain the principal eigenvector matrix.

[0163] 7. Perform dimensionality reduction by multiplying the standardized feature matrix by the principal eigenvector matrix to obtain the dimensionality-reduced feature matrix.

[0164] After final processing, the selected features are reduced to 10 dimensions, namely: standard deviation, residual, peak-to-peak value, energy, skewness, kurtosis, waveform factor, amplitude factor, impact factor, and margin factor. This ten-dimensional feature space can simplify calculations while improving the accuracy of connectivity status monitoring and diagnosis.

[0165] Please refer to Table 3 below for the filtered feature parameter data:

[0166] Table 3: Ten dimensional and dimensionless characteristic parameters of the signal after screening when the torque is 30 N·cm

[0167]

[0168] Step S06: Input the dataset into the initial signal recognition model for training to obtain the final signal recognition model, and input the test signal into the final signal recognition model to obtain the health status classification result.

[0169] It should be noted that the test signal is generated by the push rod impacting the test component with an unknown state signal according to the periodic pulse excitation signal. Since the above steps have deeply analyzed and trained vibration signals generated by a large number of known state signals, the final signal recognition model can quickly identify the test signal and obtain the state of the test component corresponding to the test signal. This state can be signals such as torque, gap, impurity content, and friction coefficient.

[0170] In some embodiments, to construct a dataset, all obtained feature parameter data need to be bound to torque results. Each known torque yields a large number of raw vibration signals. Simultaneously, impact monitoring for different torques also yields a large number of raw vibration signals. Each raw vibration signal corresponds to a known torque and a set of feature parameter data, thus constructing a large and comprehensive dataset. This constructed dataset is then input into a signal recognition model for training, resulting in the final signal recognition model. This initial signal recognition model uses the Support Vector Machine (SVM) algorithm. The classification accuracy of the final signal recognition model obtained after training with SVMs based on different kernel functions is shown in Table 4 below.

[0171] Table 4: Diagnostic accuracy of the signal formed by the superposition of IMF2-IMF4 components using SVM with different kernel functions.

[0172]

[0173] It should also be noted that the structural components mentioned in this embodiment can be structures containing bolts, studs, screws, etc., which have signals such as torque, gaps, and material purity. Since the methods of other known state signals are basically the same as those of known torque signals, this embodiment mainly uses known torque as an example for explanation. In addition, the signal to be tested is generated by the excitation signal impacting the structural component to be tested. At the same time, the signal to be tested also needs to be decomposed according to the steps in this embodiment to obtain a new combined signal. The combined signal is then input into the final signal model to obtain the torque classification result, i.e., the torque value. The accuracy rate can be as high as 97.1%. Compared with the traditional torque recognition accuracy of 86% without using noise removal technology to remove the resonance caused by the suspended fiber Bragg grating, the torque recognition accuracy of this application is greatly improved.

[0174] Please see Figure 9 The diagram shown is a schematic representation of a signal excitation and recognition system according to a second embodiment of this application. The system includes:

[0175] The finite element modal analysis module 100 is used to build a finite element simulation analysis model of the structure under test and perform modal analysis on the finite element simulation analysis model of the structure under test to obtain the natural frequency and mode of the structure under test, thereby determining the assembly method between the suspended fiber Bragg grating sensor and the structure under test, and obtaining the resonant frequency of the suspended fiber Bragg grating sensor according to the assembly method between the suspended fiber Bragg grating sensor and the structure under test.

[0176] Furthermore, in some embodiments, the finite element modal analysis module 100 further includes:

[0177] The structural analysis unit is used to determine the adhesive point position and the fiber Bragg grating position based on the modal analysis results, and to obtain the adhesive length, the distance from the adhesive point to the fiber Bragg grating, and the distance from the fiber Bragg grating to the pigtail based on the adhesive point position and the fiber Bragg grating position.

[0178] The total length calculation unit is used to calculate the total length L based on the distance from the patch to the fiber Bragg grating, the distance from the fiber Bragg grating to the pigtail, and the grating length of the fiber Bragg grating, and to calculate the resonant frequency of the suspended fiber Bragg grating sensor based on the total length L.

[0179] The resonant frequency of the suspended fiber Bragg grating sensor is calculated using the following formula:

[0180]

[0181] Where f0 represents the resonant frequency of the suspended fiber Bragg grating sensor, and C represents the velocity of sound waves in the fiber.

[0182] The excitation signal generation module 200 is used to determine the excitation period according to the natural frequency, generate a periodic pulse excitation signal by the exciter according to the excitation period, and generate an original vibration signal by impacting the test structure with a known state signal through the push rod according to the periodic pulse excitation signal. The known state signal includes at least torque signal, crack signal, friction coefficient signal, and material purity signal.

[0183] The excitation signal is obtained according to the following formula:

[0184]

[0185] in, Indicates the excitation signal. Represents the unit impulse function. N is an integer, A represents the impulse intensity, t represents time, and T represents the excitation period.

[0186] The vibration signal decomposition module 300 is used to decompose and extract the original vibration signal using the empirical mode decomposition algorithm to obtain multi-order IMF components, and to calculate the cross-correlation coefficient between the time domain signal of each IMF component and the original vibration signal.

[0187] Furthermore, in some embodiments of this application, the vibration signal decomposition module 300 further includes:

[0188] The decomposition execution unit is used to decompose the original vibration signal into two parts: local smoothing and local oscillation. The extreme points of the local smoothing part are extracted to obtain a set of local extreme point sequences. The local extreme point sequences are then interpolated and fitted to obtain a set of local smoothing functions.

[0189] The oscillation function acquisition unit is used to obtain a set of local oscillation functions based on the local smoothing function and the original vibration signal, and to obtain the first-order IMF component based on the local oscillation functions;

[0190] The repeated decomposition unit is used to obtain the remaining components after the first decomposition based on the first-order IMF component and the original vibration signal, and to repeatedly decompose and extract the extreme points of the remaining components after the first decomposition to obtain the second-order IMF component.

[0191] Repeat the above steps until the k-th order IMF component is obtained;

[0192] The cross-correlation coefficient calculation unit is used to obtain the cross-correlation coefficient according to the following formula:

[0193]

[0194] The mean of the time-domain signal of all sampled k-th order IMF components is obtained using the following formula:

[0195]

[0196] The mean of the time-domain signal of all sampled original vibration signals is obtained using the following formula:

[0197]

[0198] Where r represents the cross-correlation coefficient between the time-domain signal of the k-th IMF component and the time-domain signal of the original vibration signal, i represents the i-th sampling point, and m represents the total number of sampling points. This represents the time-domain signal of the k-th order IMF component at the i-th sampling point. Let represent the mean of the time-domain signal of all sampled k-th order IMF components. This represents the mean of the time-domain signal of all sampled original vibration signals. This represents the time-domain signal of the original vibration signal at the i-th sampling point.

[0199] The signal denoising module 400 is used to filter out at least one target IMF component from all IMF components according to the cross-correlation coefficient, and to denoise the target IMF component with suspended fiber Bragg grating resonance noise according to the resonance frequency of the suspended fiber Bragg grating sensor. After denoising, all target IMF components are superimposed to obtain a combined signal.

[0200] Furthermore, in some embodiments, the signal denoising module 400 further includes:

[0201] The IMF component filtering unit is used to determine whether the cross-correlation coefficient corresponding to any IMF component is greater than a first preset threshold, so that all IMF components that are greater than the first preset threshold are taken as target IMF components.

[0202] The denoising execution unit is used to determine the filtering frequency range based on the resonant frequency of the suspended fiber Bragg grating sensor, so as to denoise the target IMF component with resonant noise according to the filtering frequency range.

[0203] The dataset construction module 500 is used to extract dimensional parameters and dimensionless parameters from the combined signal, and to filter the dimensional parameters and the dimensionless parameters to obtain feature parameter data, so as to construct a dataset based on the feature parameter data.

[0204] Principal component analysis algorithm is used to reduce the dimensionality of all dimensional and dimensionless parameters to remove redundant variables and obtain characteristic parameter data;

[0205] The recognition result output module 600 is used to input the dataset into the initial signal recognition model for training to obtain the final signal recognition model, and input the signal to be tested into the final signal recognition model to obtain the health status classification result.

[0206] This application also proposes a storage medium on which one or more programs are stored, which, when executed by a processor, implement the above-described signal excitation and recognition methods.

[0207] In another aspect, this application also proposes a computer device, including a memory and a processor, wherein the memory is used to store computer programs, and the processor is used to execute the computer programs stored in the memory to implement the above-described signal excitation and recognition method.

[0208] Those skilled in the art will understand that the logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can mean any means that can contain stored, communicated, propagated, or transmitted programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0209] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0210] It should be understood that various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0211] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0212] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these modifications and improvements all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A signal excitation and recognition method, characterized in that, The method includes: A finite element simulation analysis model of the structure under test is constructed, and modal analysis is performed on the finite element simulation analysis model of the structure under test to obtain the natural frequencies and modes of the structure under test. This determines the assembly method between the suspended fiber Bragg grating sensor and the structure under test, and the resonant frequency of the suspended fiber Bragg grating sensor is obtained based on the assembly method. This includes: determining the adhesive point position and the fiber Bragg grating position based on the modal analysis results; obtaining the adhesive length, the distance from the adhesive point to the fiber Bragg grating, and the distance from the fiber Bragg grating to the pigtail based on the adhesive point position and the fiber Bragg grating position; calculating the total length L based on the distance from the adhesive point to the fiber Bragg grating, the distance from the fiber Bragg grating to the pigtail, and the grating length of the fiber Bragg grating; and calculating the resonant frequency of the suspended fiber Bragg grating sensor based on the total length L. The excitation period is determined based on the natural frequency. A periodic pulse excitation signal is generated by a vibrator according to the excitation period. This signal is then used to strike the test structure with a known state signal via a push rod, generating an initial vibration signal. The known state signal includes at least torque, crack, friction coefficient, and material purity signals. The excitation signal is obtained according to the following formula: ,in, Indicates the excitation signal. Represents the unit impulse function. N is an integer, A represents the impulse intensity, t represents the time, and T represents the excitation period; the step of calculating the resonant frequency of the suspended fiber Bragg grating sensor based on the total length L includes: calculating the resonant frequency of the suspended fiber Bragg grating sensor according to the following formula: Where f0 represents the resonant frequency of the suspended fiber Bragg grating sensor, and C represents the velocity of sound waves in the fiber. The original vibration signal is decomposed and extracted using an empirical mode decomposition algorithm to obtain multi-order IMF components. This includes: decomposing the original vibration signal into two parts: local smoothing and local oscillation; extracting extreme points from the locally smoothed part to obtain a sequence of local extreme points; interpolating and fitting the sequence of local extreme points to obtain a set of local smoothing functions; obtaining a set of local oscillation functions based on the local smoothing functions and the original vibration signal; obtaining the first-order IMF component based on the local oscillation functions; obtaining the remaining components after the first decomposition based on the first-order IMF component and the original vibration signal; and repeating the decomposition and extreme point extraction of the remaining components after the first decomposition to obtain the second-order IMF component; repeating the above steps until the k-th order IMF component is obtained. And calculate the cross-correlation coefficients between the time-domain signals of each IMF component and the original vibration signal, including obtaining the cross-correlation coefficients according to the following formula: The mean of the time-domain signal of all sampled k-th order IMF components is obtained using the following formula: The mean of the time-domain signal of all sampled original vibration signals is obtained according to the following formula: Where r represents the cross-correlation coefficient between the time-domain signal of the k-th IMF component and the time-domain signal of the original vibration signal, i represents the i-th sampling point, and m represents the total number of sampling points. This represents the time-domain signal of the i-th sampling point of the k-th IMF component. Let represent the mean of the time-domain signal of all sampled k-th order IMF components. This represents the mean of the time-domain signal of all sampled original vibration signals. This represents the time-domain signal of the original vibration signal at the i-th sampling point; At least one target IMF component is selected from all IMF components based on the cross-correlation coefficient, and the target IMF component with suspended fiber Bragg grating resonance noise is denoised based on the resonant frequency of the suspended fiber Bragg grating sensor. After denoising, all target IMF components are superimposed to obtain a combined signal. Dimensional and dimensionless parameters are extracted from the combined signal, and the dimensional and dimensionless parameters are filtered to obtain feature parameter data, so as to construct a dataset based on the feature parameter data. The dataset is input into the initial signal recognition model for training to obtain the final signal recognition model. The test signal is then input into the final signal recognition model to obtain the health status classification result.

2. The signal excitation and recognition method according to claim 1, characterized in that, The step of obtaining a set of local oscillation functions based on the local smoothing function and the original vibration signal, and obtaining the first-order IMF component based on the local oscillation functions, includes: The local oscillation function can be obtained using the following formula: in, Represents a local oscillation function. Represents the original vibration signal. Represents a local smoothing function; Determine whether the local oscillation function satisfies the preset IMF component conditions, which include the number of extreme points in the local oscillation function being equal to the number of zero crossover points and the average value of the envelope function being zero; If the local oscillation function does not meet the preset IMF component conditions, the obtained local oscillation function is used as a new vibration signal, and a new local oscillation function is calculated again based on the new vibration signal until the new local oscillation function meets the preset IMF component conditions. Then, the local oscillation function that meets the preset IMF component conditions is output as the first-order IMF component. If the local oscillation function satisfies the preset IMF component conditions, then the local oscillation function will be... As a first-order IMF component; The k-th order IMF component is obtained using the following formula: Where n represents the number of decompositions. Represents the k-th order IMF component. This represents the remaining components after the nth decomposition.

3. The signal excitation and recognition method according to claim 1, characterized in that, The steps of selecting at least one target IMF component from all IMF components based on the cross-correlation coefficient, denoising the target IMF component with suspended fiber Bragg grating resonance noise based on the resonant frequency of the suspended fiber Bragg grating sensor, and then superimposing all the target IMF components to obtain the combined signal include: Determine whether the cross-correlation coefficient corresponding to any IMF component of any order is greater than a first preset threshold, so that all IMF components that are greater than the first preset threshold are taken as target IMF components. The filtering frequency range is determined based on the resonant frequency of the suspended fiber Bragg grating sensor, so as to denoise the target IMF component with suspended fiber Bragg grating resonant noise according to the filtering frequency range.

4. The signal excitation and recognition method according to claim 1, characterized in that, The step of extracting dimensional parameters and dimensionless parameters from the combined signal, and filtering the dimensional parameters and dimensionless parameters to obtain feature parameter data includes: The dimensional parameters include mean, standard deviation, maximum, minimum, residual, peak-to-peak value, and energy; the dimensionless parameters include skewness, kurtosis, waveform factor, amplitude factor, impact factor, and margin factor. Principal component analysis (PCA) was used to reduce the dimensionality of all dimensional and dimensionless parameters to eliminate redundant variables and obtain characteristic parameter data.

5. A signal excitation and recognition system, characterized in that, The system is applied to the signal excitation and recognition method as described in claim 1, the system comprising: The finite element modal analysis module is used to build a finite element simulation analysis model of the structure under test and perform modal analysis on the finite element simulation analysis model of the structure under test to obtain the natural frequency and mode of the structure under test, thereby determining the assembly method between the suspended fiber Bragg grating sensor and the structure under test, and obtaining the resonant frequency of the suspended fiber Bragg grating sensor according to the assembly method between the suspended fiber Bragg grating sensor and the structure under test. The excitation signal generation module is used to determine the excitation period according to the natural frequency, generate a periodic pulse excitation signal by the exciter according to the excitation period, and generate an original vibration signal by impacting the test structure with a known state signal through the push rod according to the periodic pulse excitation signal. The known state signal includes at least torque signal, crack signal, friction coefficient signal, and material purity signal. The vibration signal decomposition module is used to decompose and extract the original vibration signal using the empirical mode decomposition algorithm to obtain multi-order IMF components, and to calculate the cross-correlation coefficient between the time domain signal of each IMF component and the original vibration signal. The signal denoising module is used to filter at least one target IMF component from all IMF components according to the cross-correlation coefficient, and to denoise the target IMF component with suspended fiber Bragg grating resonance noise according to the resonance frequency of the suspended fiber Bragg grating sensor. After denoising, all target IMF components are superimposed to obtain a combined signal. The dataset construction module is used to extract dimensional parameters and dimensionless parameters from the combined signal, and to filter the dimensional parameters and the dimensionless parameters to obtain feature parameter data, so as to construct a dataset based on the feature parameter data. The recognition result output module is used to input the dataset into the initial signal recognition model for training to obtain the final signal recognition model, and input the signal to be tested into the final signal recognition model to obtain the health status classification result.

6. A computer device, characterized in that, The computer device includes a memory and a processor, wherein: The memory is used to store computer programs; When the processor executes the computer program stored in the memory, it implements the signal excitation and recognition method according to any one of claims 1-4.

Images