Ship structure crack positioning and feature reconstruction method and system

Abstract

The invention discloses a ship structure crack positioning and feature reconstruction method and system. The method comprises the steps that a sensor array is arranged in a to-be-detected area, guided wave signals are emitted and collected, domain transformation, mode screening and matched frequency window filtering are conducted on original signals, and enhanced signal data are obtained; carrying out multi-mode feature extraction on the enhanced signal, constructing a multi-dimensional feature vector containing mode features and a coupling relation thereof, and realizing initial crack positioning by combining sensor position information; time inversion and wave field back propagation are further carried out on the enhanced signal, an energy field is constructed, accurate crack positioning is realized according to an energy field extreme value, and crack direction and size information is extracted; according to the method, the signal-to-noise ratio, the positioning precision and the stability and quality of time reversal imaging can be remarkably improved.

Description

Technical Field

[0001] This invention belongs to the field of structural health monitoring technology, and in particular relates to a method and system for locating and reconstructing features of cracks in ship structures. Background Technology

[0002] During long-term operation, the steel structure of a ship is subjected to factors such as wave impact, fatigue load, and corrosion wear, resulting in cracks of varying sizes. If the location and development trend of these cracks are not detected in time, it may lead to a decrease in structural strength or even a major accident.

[0003] In existing technologies, commonly used ultrasonic testing methods for ship structures mainly include conventional ultrasonic straight probe testing and phased array ultrasonic testing. These methods are effective in localized areas such as flat plates and welds, but conventional ultrasonic testing is limited to short-range spot inspections and is difficult to perform comprehensive inspections of large ship hulls. Ship structures include components such as stiffeners, ribs, and bulkheads, and reflections and mode conversions are unavoidable, complicating waveform signals and making accurate localization difficult using conventional methods. Different crack lengths, inclination angles, and opening degrees significantly affect signal characteristics, leading to low localization stability. To address problems in large-scale and complex operating conditions, ultrasonic guided wave testing has become a research hotspot. Guided waves can propagate over long distances within structures, facilitating large-scale monitoring. However, existing guided wave methods suffer from multiple guided wave modes and complex propagation paths; multi-path superposition makes it difficult to analyze reflected signals; and crack location and size cannot be accurately inverted from a single reflected wave. Summary of the Invention

[0004] Purpose of the invention: The purpose of this invention is to provide a method and system for locating and reconstructing crack features in ship structures that can improve the accuracy of crack location.

[0005] Technical solution: The present invention provides a method for locating and reconstructing features of cracks in ship structures, comprising:

[0006] A sensor array is arranged in the area to be tested of the object being tested, and the position information of each sensor in the sensor array is recorded;

[0007] The sensor array emits guided wave signals and acquires response signals to obtain raw signal data;

[0008] The original signal data is subjected to domain transformation and screening of several target modes, and then filtered using a frequency window that matches each of the screened target modes to obtain enhanced signal data.

[0009] Multi-mode feature extraction is performed on the enhanced signal data to construct a multi-dimensional feature vector containing the features of each target mode and the coupling features between modes; the features of each target mode include the peak amplitude, time of arrival, and group velocity change rate of each target mode, and the coupling features between modes include the phase difference and mode conversion energy ratio between each target mode;

[0010] Based on the multidimensional feature vector and the position information of the sensor, the preliminary location result of the crack is obtained.

[0011] The enhanced signal data is subjected to time reversal processing and wave field back propagation calculation to construct an energy field. The precise location result of the crack is obtained based on the extreme points of the energy field. The result is compared and verified with the preliminary location result to confirm the final location result.

[0012] Based on the spatial distribution characteristics of the energy field around the precise positioning result, the direction and size information of the crack are obtained, and feature reconstruction is completed.

[0013] Further, the step of performing domain transformation and target mode filtering on the original signal data, and then filtering it using a frequency window matching the selected target mode to obtain enhanced signal data, includes:

[0014] The original signal data is subjected to Fourier transform along the time and spatial dimensions to obtain the frequency-wavenumber domain signal distribution;

[0015] The frequency-wavenumber domain signal distribution is compared and matched with the theoretical dispersion curve of the object under test to identify each guided wave mode and its energy distribution. Based on the displacement field distribution characteristics, scattering energy ratio and mode separation degree of each guided wave mode, the target mode is selected.

[0016] Based on the dispersion characteristics of each target mode, the concentrated frequency range corresponding to each target mode is determined, and a bandpass frequency window is constructed within each concentrated frequency range; the concentrated frequency range is specifically the range in which the energy corresponding to the current target mode accounts for a proportion of the total energy of the entire frequency band greater than 0.7, and the ratio of the distance between the wavenumber of the current target mode and the wavenumber of the adjacent mode is greater than 0.15.

[0017] The original signal data is filtered using each of the aforementioned bandpass frequency windows to obtain the signal components corresponding to each target mode. The signal components corresponding to each target mode constitute the enhanced signal data.

[0018] Furthermore, the selection of target modes based on the displacement field distribution characteristics, scattering energy ratio, and mode separation degree of each guided wave mode includes:

[0019] In the frequency-wavenumber domain signal distribution, the ratio of the energy of each guided wave mode to the energy of the direct wave within the time window corresponding to the crack reflection wave is calculated, and the ratio is defined as the scattered energy ratio. Guided wave modes with a scattered energy ratio higher than a preset energy ratio threshold are selected as candidate target modes.

[0020] In the frequency-wavenumber domain, the mode separation degree between each candidate target mode and other modes is evaluated. The mode separation degree is defined as the ratio of the wavenumber difference between adjacent modes at the same frequency to the average wavenumber. Modes with a mode separation degree greater than a preset separation threshold are selected as target modes.

[0021] Further, the step of performing multi-mode feature extraction on the enhanced signal data to construct a multi-dimensional feature vector containing features of each target mode and coupling features between modes includes:

[0022] Envelope extraction is performed on the signal components of each target mode in the enhanced signal data to obtain the peak amplitude and arrival time of each target mode;

[0023] The measured group velocity of each target mode is calculated based on the distance between signal acquisition points and the arrival time of the direct wave. The measured group velocity is then compared with the theoretical group velocity to obtain the rate of change of the group velocity of each target mode.

[0024] Instantaneous phase extraction is performed on the signal components of each target mode, and the phase difference between each target mode at the crack reflection wave is calculated.

[0025] Calculate the ratio of mode conversion energy to total reflected energy among the target modes to obtain the mode conversion energy ratio;

[0026] The peak amplitude, arrival time, group velocity change rate, phase difference, and mode transition energy ratio of each target mode are integrated to construct the multidimensional feature vector.

[0027] Further, the step of obtaining the preliminary crack location result based on the multidimensional feature vector and the sensor's position information includes:

[0028] The theoretical group velocity of each target mode is corrected based on the group velocity change rate in the multidimensional feature vector to obtain the corrected group velocity.

[0029] Based on the arrival time of each target mode in the multidimensional feature vector and the corrected group velocity, the propagation path length of the guided wave from the sensor's signal emission point to the receiving point via crack reflection is calculated.

[0030] Based on the propagation path length and the position information of each sensor, an elliptical equation is established with the crack location coordinates as unknowns;

[0031] Based on the phase difference between each target pattern in the multidimensional feature vector, data whose phase difference exceeds a preset phase difference threshold are removed.

[0032] Based on the mode conversion energy ratio in the multidimensional feature vector, mode combinations with a mode conversion energy ratio higher than a preset energy ratio threshold are selected to participate in the localization calculation.

[0033] An overdetermined system of equations is established using multiple sets of propagation path lengths. Weighting coefficients are constructed based on the peak amplitude of each target mode in the multidimensional feature vector. The weighting coefficients are obtained by normalizing the peak amplitude of each propagation path by dividing it by the maximum value of the peak amplitude of all propagation paths. The values ​​range from 0 to 1. The overdetermined system of equations is solved using a weighted optimization algorithm to obtain the preliminary location results of the crack.

[0034] Furthermore, the enhanced signal data undergoes time-reversal processing and wavefield backpropagation calculation to construct an energy field. Based on the extreme points of the energy field, the precise location of the crack is obtained. This is then compared and verified with the preliminary location results to confirm the final location result, including:

[0035] The sampled value sequences of each signal channel in the enhanced signal data are arranged in reverse order along the time axis to obtain the time-reversed signal;

[0036] The time-reversed signal is used as a boundary condition input to the wave equation numerical model to calculate the backward propagation process of the guided wave in the measured object and obtain the displacement field distribution at each time.

[0037] Cross-correlation processing is performed on the backpropagation sound fields of each sensor channel, and the product of the sound fields of different channels is integrated along the time axis to construct a cross-energy field as the energy field.

[0038] The coordinates of the point with the maximum energy value are extracted from the energy field to obtain the precise location of the crack.

[0039] The precise positioning result is compared and verified with the preliminary positioning result. If the deviation between the two is within a preset threshold range, the precise positioning result is confirmed as the final positioning result.

[0040] Furthermore, obtaining the direction and size information of the crack based on the spatial distribution characteristics of the energy field around the precise positioning result includes:

[0041] Using the precise positioning result as the center, extract the energy contour lines of the energy field within a local area of ​​a preset range;

[0042] Principal component analysis is performed on the set of coordinate points within the energy contour lines to obtain crack direction information based on the direction of the principal eigenvectors.

[0043] Energy profiles are extracted along the direction corresponding to the crack direction information, and crack size information is obtained based on the width range of the energy profiles and a preset calibration relationship.

[0044] Further, the step of performing principal component analysis on the set of coordinate points within the energy contour lines, obtaining crack direction information based on the direction of the principal eigenvectors, and completing feature reconstruction includes:

[0045] Extract the coordinates of all points within the energy contour lines and calculate the centroid of the coordinate set;

[0046] The coordinate set is decentered using the centroid as the origin to construct a covariance matrix;

[0047] The covariance matrix is ​​decomposed into eigenvalues ​​to obtain eigenvalues ​​and corresponding eigenvectors.

[0048] The eigenvector corresponding to the largest eigenvalue is determined as the main eigenvector, and the crack direction information is obtained based on the angle between the main eigenvector and the reference axis.

[0049] Further, the step of extracting the energy profile along the direction corresponding to the crack direction information, and obtaining crack size information based on the width range of the energy profile and a preset calibration relationship, includes:

[0050] Along the direction corresponding to the crack direction information, take a line through the precise positioning result in the energy field to obtain a one-dimensional energy profile;

[0051] On the one-dimensional energy profile, two location points are determined where the energy value drops to a preset percentage of the peak value, and the distance between the two location points is calculated as the expansion range.

[0052] The width expansion range is converted into crack size information based on the calibration coefficients obtained in advance through calibration of crack samples of known size.

[0053] Based on the same inventive concept, the present invention also provides a system for locating and reconstructing cracks in ship structures, comprising:

[0054] A sensor array is arranged in the area to be detected of the object being measured, and is used to emit guided wave signals and acquire response signals;

[0055] The signal acquisition module is used to record the position information of each sensor in the sensor array and to acquire the response signals collected by the sensor array to form raw signal data;

[0056] The signal enhancement module is used to perform Fourier transform on the original signal data along the time and space dimensions to obtain the frequency-wavenumber domain signal distribution, compare and match the frequency-wavenumber domain signal distribution with the theoretical dispersion curve of the object under test to identify and filter target modes, construct a bandpass frequency window according to the dispersion characteristics of each target mode, and use the bandpass frequency window for filtering to obtain enhanced signal data.

[0057] The feature extraction module is used to perform multi-mode feature extraction on the enhanced signal data, obtain the peak amplitude, arrival time, group velocity change rate of each target mode, as well as the phase difference and mode conversion energy ratio between each target mode, and construct a multi-dimensional feature vector containing the features of each target mode and the coupling features between modes.

[0058] The preliminary positioning module is used to obtain the preliminary positioning result of the crack by establishing an elliptic equation system and solving it using a weighted optimization algorithm based on the arrival time, corrected group velocity, peak amplitude, phase difference, and mode conversion energy ratio in the multidimensional feature vector and the position information of the sensor.

[0059] The precise positioning module is used to perform time reversal processing and wave field back propagation calculation on the enhanced signal data, construct an energy field through cross-correlation processing, obtain the precise positioning result of the crack based on the extreme points of the energy field, compare and verify with the preliminary positioning result, and confirm the final positioning result.

[0060] The feature reconstruction module is used to obtain the direction and size information of the crack through principal component analysis and energy profile broadening calculation based on the spatial distribution characteristics of the energy field around the precise positioning result, and to complete the feature reconstruction.

[0061] Beneficial effects: Compared with existing technologies, this invention effectively solves the signal aliasing problem caused by multipath interference in complex ship structures through mode selection and frequency window matching preprocessing methods; the constructed dual-channel guided wave reflection feature vector integrates amplitude, time, velocity, and modal coupling information of the two modes, with rich and complementary information dimensions; it achieves effective separation of the fundamental antisymmetric mode and the fundamental symmetric mode, which significantly improves the signal-to-noise ratio and crack location accuracy, and improves the stability and imaging quality of time-inversion focusing; this invention also establishes a quantitative correlation model between the time-inversion focusing energy field and crack geometric features, realizing non-contact reconstruction of crack length and tilt angle; this invention can simultaneously output complete crack geometric features, providing more comprehensive data support for structural safety assessment. Attached Figure Description

[0062] Figure 1 This is a flowchart illustrating the overall method of an embodiment of the present invention;

[0063] Figure 2This is a flowchart of the feature reconstruction method according to an embodiment of the present invention;

[0064] Figure 3 A flowchart illustrating the method for constructing enhanced signal data according to an embodiment of the present invention;

[0065] Figure 4 A flowchart illustrating the method for constructing multidimensional feature vectors according to an embodiment of the present invention;

[0066] Figure 5 This is a flowchart of the method for solving the preliminary localization problem according to an embodiment of the present invention;

[0067] Figure 6 This is a flowchart illustrating the method for confirming the final location result in an embodiment of the present invention.

[0068] Figure 7 This is a flowchart of a method for obtaining crack direction information according to an embodiment of the present invention;

[0069] Figure 8 This is a flowchart illustrating the method for obtaining crack size information according to an embodiment of the present invention.

[0070] Figure 9 This is a system structure diagram of an embodiment of the present invention. Detailed Implementation

[0071] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0072] As attached Figure 1 As shown, the ship structure crack location and feature reconstruction method of this embodiment includes:

[0073] Step 100: Arrange a sensor array in the area to be detected of the object being tested, and record the position information of each sensor in the sensor array;

[0074] Step 200: Transmit guided wave signals through the sensor array and acquire response signals to obtain raw signal data;

[0075] Step 300: Perform domain transformation and screening of several target modes on the original signal data, and use frequency windows that match the screened target modes for filtering to obtain enhanced signal data;

[0076] Step 400: Perform multi-mode feature extraction on the enhanced signal data to construct a multi-dimensional feature vector containing the features of each target mode and the coupling features between modes; the features of each target mode include the peak amplitude, time of arrival, and group velocity change rate of each target mode, and the coupling features between modes include the phase difference and mode conversion energy ratio between each target mode;

[0077] Step 500: Based on the multidimensional feature vector and the position information of the sensor, the preliminary location result of the crack is obtained;

[0078] Step 600: Perform time reversal processing and wave field backpropagation calculation on the enhanced signal data to construct an energy field, and obtain the precise location result of the crack based on the extreme points of the energy field. Compare and verify with the preliminary location result to confirm the final location result.

[0079] Step 700: Based on the spatial distribution characteristics of the energy field around the precise positioning result, obtain the direction and size information of the crack, and complete the feature reconstruction.

[0080] Specifically, in step 100, it should be understood that, based on the structural form of the object under test and the testing requirements, a sensor array is arranged in a preset geometric layout within the area to be tested of the object. The sensor array includes multiple piezoelectric sensors, each of which can act as a guided wave excitation source to emit guided wave signals and as a receiver to collect guided wave response signals. Here, the object under test refers to the ship structural component to be inspected for cracks. Position information refers to the spatial coordinate data of each sensor in the coordinate system of the object under test.

[0081] In step 200, it should be understood that the guided wave signal propagates in the object under test and is reflected and scattered when it encounters defects such as cracks. The response signals after the guided wave signal propagates are collected by other sensors in the sensor array, and the response signals collected by each sensor are digitized and stored to obtain the raw signal data.

[0082] In one embodiment of this application, as Figure 3 As shown, in step 300, a bandpass frequency window is constructed based on the dispersion characteristics of each target mode. The original signal data is then filtered using the bandpass frequency window to obtain the signal components corresponding to each target mode. These signal components constitute the enhanced signal data. This includes the following steps:

[0083] Step 301: Perform Fourier transform on the original signal data along the time and spatial dimensions respectively to obtain the frequency-wavenumber domain signal distribution.

[0084] Specifically, for the original signal matrix Perform a two-dimensional Fourier transform to convert the time-space domain signal into a frequency-wavenumber domain signal distribution. The specific process of the two-dimensional Fourier transform is as follows: First, a fast Fourier transform is performed on the signal along the time dimension to obtain a frequency domain representation; then, a fast Fourier transform is performed along the spatial dimension, i.e., the direction of the sensor array, to obtain a wavenumber domain representation. The transformed signal... It can clearly show the distribution characteristics of each guided wave mode on the frequency-wavenumber plane.

[0085] Step 302: Compare and match the frequency-wavenumber domain signal distribution with the theoretical dispersion curve of the object under test, identify each guided wave mode and its energy distribution, and select the guided wave mode that is sensitive to crack scattering as the target mode based on the displacement field distribution characteristics, scattering energy ratio and mode separation degree of each guided wave mode.

[0086] Specifically, based on the frequency-wavenumber domain signal distribution The theoretical dispersion curve of the guided wave propagating in the plate structure is obtained by combining the theoretical solution of the guided wave dispersion equation. The frequency-wavenumber domain signal distribution is superimposed and compared with the theoretical dispersion curve, and the measured frequency-wavenumber distribution is then compared. By comparing and matching the signal with the theoretical dispersion curve, the actual guided wave modes and their energy distribution in the signal are identified. This step also obtains the phase velocity of each mode at different frequencies. Group speed Data. Among the identified waveguide modes, based on the sensitivity of each waveguide mode to scattering of crack-like defects, waveguide modes sensitive to crack scattering are selected as target modes. Target modes include, but are not limited to, the basic antisymmetric mode A0 and the basic symmetric mode S0.

[0087] Furthermore, the displacement field distribution characteristics of each waveguide mode in the plate thickness direction were analyzed, and modes with larger displacement components within the crack depth range were preferentially selected. For surface cracks, modes with larger surface displacement components were preferred; for through-cracks, modes with significant displacement components throughout the entire plate thickness were preferred. The out-of-plane displacement component of the fundamental antisymmetric mode A0 is larger at the plate surface, making it highly sensitive to surface-opening cracks; the in-plane displacement component of the fundamental symmetric mode S0 is larger at the plate mid-surface, making it highly sensitive to through-cracks.

[0088] In the frequency-wavenumber domain signal distribution, the ratio of the energy of each guided wave mode within the time window corresponding to the crack reflection wave to the energy of the direct wave is calculated, and this ratio is defined as the scattered energy ratio. The larger the scattered energy ratio, the stronger the interaction between the mode and the crack, and the more sensitive it is to crack scattering. Guided wave modes with scattered energy ratios higher than a preset threshold are selected as candidate target modes. The preset threshold can be set according to the structural characteristics of the object under test and the detection requirements, with a typical value range of 0.05 to 0.2.

[0089] In the frequency-wavenumber domain, the separation degree between each candidate target mode and other modes is evaluated. Modes with sufficient wavenumber intervals from other modes within the operating frequency range are selected to ensure that the filtering process can effectively separate the target mode signals. The mode separation degree is defined as the ratio of the wavenumber difference between adjacent modes at the same frequency to the average wavenumber. Modes with a mode separation degree greater than a preset separation threshold are selected, with typical separation threshold values ​​ranging from 0.1 to 0.3.

[0090] Step 303: Based on the dispersion characteristics of each target mode, determine the concentrated frequency range corresponding to each target mode, and construct a bandpass frequency window within each concentrated frequency range.

[0091] Specifically, matching bandpass frequency windows are constructed for target modes A0 and S0 respectively. and Based on the obtained dispersion curves, the frequency range in which the A0 mode has concentrated energy and relatively weak dispersion was determined; within this range, the center frequency that maximizes the wavenumber domain separation between the A0 mode and other modes, especially the S0 mode, was selected. Determine the frequency window bandwidth based on the signal bandwidth and frequency resolution requirements. Frequency window Use Hanning or Gaussian windows to reduce spectral leakage.

[0092] S0 mode frequency window The design process is similar to that of A0: determine the frequency range where the S0 mode has concentrated energy and high separation from the A0 mode; select the center frequency. and bandwidth Constructing window functions .

[0093] It should be noted that the center frequency and bandwidth of the A0 mode frequency window and the S0 mode frequency window are usually different. In typical ship steel plate structures, the center frequency of the A0 mode frequency window is usually lower than that of the S0 mode. For example, the A0 mode can select the 80 to 120 kHz frequency band, while the S0 mode can select the 140 to 200 kHz frequency band. The specific parameters are adjusted according to the actual structural thickness.

[0094] Energy concentration means that the energy of the target mode in this frequency range accounts for more than 0.7% of the total energy of the entire frequency band; high separation means that the ratio of the wavenumber spacing between the target mode and adjacent modes in this frequency range to the wavenumber of the target mode is greater than 0.15.

[0095] Step 304: Filter the original signal data using each bandpass frequency window to obtain the signal components corresponding to each target mode. The signal components corresponding to each target mode constitute the enhanced signal data.

[0096] Specifically, for the original signal matrix Apply frequency windows respectively and Filtering is performed. The filtering process is implemented in the frequency domain: firstly, ... The spectrum is obtained by performing a Fourier transform. Then, respectively with frequency windows and The components are multiplied, and then an inverse Fourier transform is performed to return to the time domain. Filtering removes higher-order modes, such as A1 and S1, as well as interference modes generated by reflections from structural boundaries, yielding an enhanced signal matrix. .

[0097] Enhanced signal matrix It contains two components: A0 mode signal components obtained by filtering and through S0 mode signal components obtained by filtering These two components retain the main information related to crack scattering in their respective modes, while significantly suppressing multipath interference and mode aliasing noise.

[0098] In one embodiment of this application, as Figure 4 As shown, step 400 involves multi-mode feature extraction of the enhanced signal data to construct a multi-dimensional feature vector containing features of each target mode and inter-mode coupling features. It should be understood that multi-mode feature extraction refers to the process of extracting feature parameters reflecting crack scattering characteristics from the signal components of multiple guided wave modes. This includes the following steps:

[0099] Step 401: Extract the envelope of the signal components of each target mode in the enhanced signal data to obtain the peak amplitude and arrival time of each target mode.

[0100] Specifically, peak amplitude This represents the maximum amplitude of the crack-reflected wave. The extraction method is as follows: [The text abruptly ends here, likely due to an incomplete sentence or missing information.] The Hilbert transform is performed to obtain the analytic signal, the signal envelope is calculated, and the maximum value of the envelope within the time window corresponding to the crack reflection wave is extracted as the peak amplitude. This parameter reflects the scattering intensity of the A0 mode guided wave by the crack and is related to the crack depth and length. Arrival time This represents the time corresponding to the peak value of the crack reflection wave. The extraction method is to locate the peak amplitude on the signal envelope. The corresponding time point, that time point is the arrival time. .

[0101] Peak amplitude This represents the maximum amplitude of the crack reflection wave in mode S0. Extraction method and... Similarly, for A Hilbert transform is performed to obtain the envelope, and the maximum envelope value within the corresponding time window of the reflected wave is extracted. The peak amplitude of the S0 mode is more sensitive to through-cracks. Time of arrival This indicates the time corresponding to the peak value of the S0 mode crack reflection wave. Since the group velocity of the S0 mode is usually higher than that of the A0 mode, the arrival time of the S0 mode reflection wave from a crack at the same location is... Generally earlier than The difference in arrival times between the two modes can be used to verify the consistency of crack localization results.

[0102] Step 402: Calculate the measured group velocity of each target mode based on the distance between signal acquisition points and the arrival time of the direct wave, compare the measured group velocity with the theoretical group velocity, and obtain the group velocity change rate of each target mode.

[0103] Specifically, based on the known distance between the excitation and receiving sensors in the sensor array and the arrival time of the direct wave of each target mode, the actual group velocity of each target mode propagating in the measured object is calculated, thus obtaining the measured group velocity of each target mode. The measured group velocity of each target mode is then compared with the theoretical group velocity calculated based on material parameters, and the relative deviation between the two is calculated to obtain the group velocity change rate of each target mode. This represents the change in measured group velocity relative to the theoretical group velocity in A0 mode. The calculation method is as follows: the measured group velocity is calculated based on the sensor spacing and the arrival time of the direct wave. The theoretical group velocity is obtained from the dispersion curve. Calculate the rate of change The group velocity variation rate reflects the local changes in the structural state near the crack. The crack causes a decrease in local stiffness, thus affecting the guided wave group velocity. This represents the change in measured group velocity relative to the theoretical group velocity in the S0 mode. The calculation method is the same as... Similarly, the S0 mode group velocity is sensitive to changes in plate thickness. Stress concentration at cracks can lead to changes in local material properties, which in turn affects the S0 mode group velocity.

[0104] Step 403: Perform instantaneous phase extraction on the signal components of each target mode and calculate the phase difference between each target mode at the crack reflection wave.

[0105] Specifically, for the signal components corresponding to each target mode in the enhanced signal data, Hilbert transform is applied to obtain analytical signals, and instantaneous phase information is extracted from the analytical signals. At the arrival time of the crack reflection wave of each target mode, the corresponding instantaneous phase value is read, and the difference between the instantaneous phase values ​​of different target modes is calculated to obtain the phase difference between each target mode at the crack reflection wave. This represents the phase shift between the crack reflection waves of mode A0 and mode S0. The calculation method is as follows: [The text abruptly ends here, likely due to an incomplete sentence or missing information.] and Perform a Hilbert transform to obtain the instantaneous phase, and extract the instantaneous phase value at the peak of the crack reflected wave. and Calculate the phase difference Phase difference Closely related to the crack inclination angle: when the crack is perpendicular to the direction of guided wave propagation, the phase difference between the two modes of reflected waves is small; when the crack is inclined, due to the different interaction mechanisms between the A0 and S0 modes and the crack, a significant phase difference will be generated.

[0106] Step 404: Calculate the ratio of mode conversion energy to total reflection energy among the target modes to obtain the mode conversion energy ratio.

[0107] Specifically, in the enhanced signal data, the crack reflection wave signal energy of each target mode is calculated separately. When the guided wave interacts with the crack, some energy undergoes mode conversion, that is, it changes from one guided wave mode to another. The mode conversion energy ratio is obtained by calculating the ratio of the signal energy after mode conversion to the total reflected energy of the crack reflection waves of all target modes. This represents the proportion of energy generated by mode conversion at the crack. When a guided wave encounters a crack, some A0 mode energy will be converted to S0 mode, and vice versa. The calculation method is as follows: Extract the A0 component energy from the S0 mode conversion in the signal. Extract the S0 component energy from the signal converted from A0 mode, and calculate the total conversion energy. Calculate the total reflected energy: Finally, the mode conversion energy ratio was obtained. Mode switching energy ratio Related to crack depth and opening angle: shallow cracks mainly produce reflections with less mode conversion, while through cracks produce significant mode conversion.

[0108] Step 405: Integrate the peak amplitude, arrival time, group velocity change rate, phase difference and mode conversion energy ratio of each target mode to construct a multidimensional feature vector.

[0109] Specifically, the peak amplitude, arrival time, and group velocity change rate of each target mode are used as single-mode feature parameters, and the phase difference and mode conversion energy ratio between target modes are used as inter-mode coupling feature parameters. The above single-mode feature parameters and inter-mode coupling feature parameters are integrated and arranged in a preset order to construct a multi-dimensional feature vector containing the features of each target mode and the inter-mode coupling features. .

[0110] Among them, peak amplitude is used to construct weighting coefficients during positioning, group velocity change rate is used to correct the propagation path length calculation, phase difference is used to determine the validity of crack reflection signals, and mode conversion energy ratio is used to screen mode combinations participating in positioning calculation.

[0111] In one embodiment of this application, as Figure 5 As shown, step 500, based on the multidimensional feature vector and the sensor's position information, obtains the preliminary crack location result, including the following steps:

[0112] Step 501: Correct the theoretical group velocity of each target mode based on the group velocity change rate in the multidimensional feature vector to obtain the corrected group velocity.

[0113] Specifically, the group velocity change rate of each target mode is extracted from the multidimensional feature vector, and the theoretical group velocity of each target mode is compensated and corrected based on the group velocity change rate. The formula for calculating the corrected group velocity is: the corrected group velocity equals the theoretical group velocity multiplied by the correction factor corresponding to the group velocity change rate. Through the above correction, the influence of the difference between the actual material state of the measured object and the theoretical model on the group velocity estimation is eliminated, and a more accurate corrected group velocity reflecting the actual propagation speed of the guided wave in the measured object is obtained.

[0114] Step 502: Based on the arrival time and corrected group velocity of each target mode in the multidimensional feature vector, calculate the propagation path length of the guided wave from the excitation point (the sensor's signal transmission point) through the crack reflection to the receiving point.

[0115] Specifically, based on arrival time and By combining the group velocities of each mode, the total path length of the guided wave from the excitation sensor, through the crack reflection, to the receiving sensor is calculated. For the first... The excitation sensor and the first For a sensor pair consisting of several receiving sensors, the path length in A0 mode is:

[0116]

[0117] in, For A0 mode group speed, This is the arrival time of the sensor for the measured A0 mode.

[0118] Step 503: Based on the propagation path length and the position information of each sensor, establish an elliptical equation with the crack location coordinates as unknowns.

[0119] Specifically, the spatial coordinates of each excitation and receiving point are determined based on the sensor's position information. An ellipse equation is established with the excitation and receiving points as the two foci of an ellipse, and the propagation path length as the major axis length, with the crack location coordinates as the unknowns. Assume the crack location coordinates are... The location of the excitation sensor is The location of the receiving sensor is Then the path length satisfies the equation of an ellipse:

[0120]

[0121] An overdetermined system of equations is established using arrival time data from multiple pairs of sensors, and the crack location coordinates are solved using the least squares method or iterative optimization algorithm. The elliptic equation represents the possible location of a crack on an elliptical trajectory with the excitation and receiver points as foci and the propagation path length as the major axis. Multiple elliptic equations can be established for different excitation-receiver sensor combinations and different target modes; the intersection of these multiple elliptical trajectories represents the possible location of the crack.

[0122] Step 504: Based on the phase difference between each target pattern in the multidimensional feature vector, remove abnormal data whose phase difference exceeds a preset threshold.

[0123] Specifically, the phase difference between each target mode at the crack reflection wave is extracted from the multi-dimensional feature vector, and the phase difference is compared with a preset threshold. When the phase difference corresponding to a certain sensor combination or a certain target mode exceeds the preset threshold, the data is determined to be abnormal data, indicating that the corresponding reflection signal may originate from structural boundary reflection, multiple reflections, or noise interference rather than the actual crack scattering signal. Abnormal data with phase differences exceeding the preset threshold are removed from the positioning calculation to improve the reliability and accuracy of the positioning calculation.

[0124] Step 505: Based on the mode conversion energy ratio in the multidimensional feature vector, select mode combinations with a mode conversion energy ratio higher than a preset value to participate in the localization calculation.

[0125] Specifically, the mode conversion energy ratio (MCE) between each target mode is extracted from the multidimensional feature vector, and then compared with a preset value. Mode combinations with MCEs higher than the preset value are selected for localization calculation, while those with MCEs lower than the preset value are discarded. A higher MCE indicates a significant mode coupling effect between the guided wave and the crack, and the corresponding reflected signal has a higher confidence level in crack characteristics. Including this in the localization calculation helps improve localization accuracy.

[0126] Step 506: Establish an overdetermined system of equations using multiple propagation path lengths. Construct weight coefficients based on the peak amplitude of each target mode in the multidimensional feature vector. Solve the overdetermined system of equations using a weighted optimization algorithm to obtain preliminary crack localization results. The weight coefficients are calculated by normalizing the peak amplitude of each propagation path by dividing it by the maximum value of the peak amplitudes of all propagation paths, and their values ​​range from 0 to 1. The weight coefficients corresponding to propagation paths with higher peak amplitudes are greater than or equal to 0.5, while the weight coefficients corresponding to propagation paths with lower peak amplitudes are less than 0.5.

[0127] Specifically, using the multiple propagation path lengths retained after filtering in steps 504 and 505, an overdetermined system of equations is established with the crack location coordinates as unknowns. The number of equations in the overdetermined system is greater than the number of unknowns. The peak amplitude of each target mode is extracted from the multidimensional feature vector. Weighting coefficients are constructed for each equation based on the peak amplitude; equations corresponding to propagation paths with higher peak amplitudes are assigned larger weights, while those corresponding to propagation paths with lower peak amplitudes are assigned smaller weights. The overdetermined system of equations is solved using weighted least squares or other weighted optimization algorithms to obtain the crack location coordinate estimate that minimizes the sum of squared weighted residuals, serving as the preliminary crack localization result.

[0128] In one embodiment of this application, as follows: Figure 6 As shown, step 600 involves performing time-reversal processing and wavefield backpropagation calculations on the enhanced signal data to construct an energy field. Based on the extreme points of the energy field, the precise location of the crack is obtained. This is then compared and verified with the preliminary location results to confirm the final location. The steps include the following:

[0129] Step 601: Arrange the sampled value sequences of each signal channel in the enhanced signal data in reverse order according to the time axis to obtain the time-reversed signal.

[0130] Specifically, the obtained enhanced signal matrix Process in reverse order along the time axis. Let the time series of the original signal be... The sampled value sequence is The signal after time reversal The corresponding time series remains unchanged. Time-reverse processing is performed on each signal in the enhanced signal matrix to obtain the time-reverse signal matrix. In the original signal, the first arriving direct wave component becomes the last emitted after being reversed, while the later arriving crack reflection wave component becomes the first emitted. When these reversed signals propagate in the medium, each component will propagate in the opposite direction to the original path and arrive at the crack location simultaneously, forming energy focusing.

[0131] Step 602: Input the time-reversed signal as a boundary condition into the numerical model of the wave equation, calculate the backward propagation process of the guided wave in the measured object, and obtain the displacement field distribution at each time.

[0132] Specifically, the time-reversed signal matrix The numerical model of the wave equation is used as a boundary condition to simulate the backward propagation of guided waves within the structure. The numerical model solves the elastic wave equation using either the finite difference method or the finite element method.

[0133]

[0134] in, For material density, and Let Lamé constant be . This is the displacement vector. Time-reversal signals are applied at each sensor location in the model. As displacement boundary conditions, the evolution of the wave field over time is solved throughout the computational domain to obtain the displacement field distribution at each moment. The computational domain covers the detection area enclosed by the sensor array, and the grid size should be less than one-tenth of the shortest wavelength to ensure computational accuracy. The time step is determined according to the Courant-Friedrich-Lévy condition to ensure numerical stability.

[0135] Step 603: Perform cross-correlation processing on the backpropagation sound fields of each sensor channel, integrate the product of the sound fields of different channels along the time axis, and construct a cross-energy field as the energy field.

[0136] Specifically, for A single excitation-receiver sensor generates M-channel backpropagating sound fields. The cross-energy field is defined as:

[0137]

[0138] in, The summation iterates through all sensor pair combinations, while integration is performed along the time axis. Compared to simple energy superposition imaging, cross-energy field imaging can effectively suppress spurious focal points generated by a single channel, improving imaging contrast and positioning resolution. For multipath interference in complex ship structures, cross-correlation operations can filter out incoherent noise components between different channels, further improving the imaging signal-to-noise ratio.

[0139] Step 604: Extract the coordinates of the point with the maximum energy value from the energy field to obtain the precise location result of the crack.

[0140] Specifically, the cross-energy field constructed in step 603 The analysis was performed, and the coordinates of the global maximum point were extracted as the precise location of the crack. To improve positioning accuracy, parabolic interpolation or Gaussian fitting methods are used near the global maximum point for sub-pixel-level refinement to obtain more accurate crack location coordinates.

[0141] Step 605: Accurately locate the results. Compared with the preliminary solution results of step 506 A comparative verification is performed. If the deviation between the two is within a preset threshold range, the positioning result is confirmed as valid, and the precise positioning result is the final positioning result; if the deviation exceeds the threshold, the signal quality needs to be checked and reprocessed. Time-reversal focusing imaging results. As the final coordinates for locating the crack.

[0142] In step 7, it should be understood that the orientation information refers to the tilt angle of the crack relative to the reference coordinate axis, characterizing the direction of crack propagation within the plane of the measured object. The size information refers to the length of the crack along its propagation direction, obtained through conversion between the energy profile broadening range and the calibration coefficient.

[0143] In one embodiment of this application, as Figure 2 Step 7 includes the following steps:

[0144] Step 701: Using the precise positioning result as the center, extract the energy contour lines of the energy field within a preset range of local areas.

[0145] It should be understood that the area enclosed by the energy contour lines represents the spatial distribution range of the time-reversed focused energy field at the crack location. Its shape characteristics are related to the geometry of the crack, exhibiting an elliptical distribution characteristic that is elongated along the crack extension direction.

[0146] Step 702: Perform principal component analysis on the set of coordinate points within the energy contour lines to obtain crack direction information based on the direction of the principal eigenvectors.

[0147] Step 703: Extract the energy profile along the direction corresponding to the crack direction information, and obtain the crack size information based on the width range of the energy profile and the preset calibration relationship.

[0148] Furthermore, in one embodiment of this application, as Figure 7 As shown, in step 702, principal component analysis is performed on the set of coordinate points within the energy contour lines to obtain crack orientation information based on the direction of the principal eigenvectors. This includes the following steps:

[0149] Step 702.1: Extract the coordinates of all points within the energy contour lines and calculate the centroid of the coordinate set.

[0150] Specifically, by the precise location of the crack Extract the cross-energy field in the local area surrounding the center. Energy contour lines. Let the local region be... ,in and The value is determined based on the expected crack size and imaging resolution, and is taken as 2 to 5 times the waveguide wavelength.

[0151] Within a local region, the energy field is normalized: This makes the maximum value 1. Then, contour lines for multiple energy levels are extracted. ,in This represents the energy threshold level, with a value of [value missing]. Etc. Energy contour lines Indicates that the energy value is equal to The contour lines are closed curves formed by all the points. For point defects, the contour lines tend to be circular; for linear cracks, the contour lines are elliptical, with the major axis of the ellipse aligned with the crack direction. The geometric characteristics of the fracture can be deduced by analyzing the shape of the contour lines.

[0152] Select energy threshold level ,choose (i.e., the equivalent of half-width and half-height), extract contour lines. The set of coordinates of all points inside ,in This represents the number of pixels within the contour lines.

[0153] Calculate the centroid of the coordinate set:

[0154] .

[0155] Step 702.2: Decentralize the coordinate set with the centroid as the origin to construct the covariance matrix.

[0156] Specifically, a decentralized coordinate matrix is ​​constructed, and the covariance matrix C is calculated:

[0157] .

[0158] Step 702.3: Perform eigenvalue decomposition on the covariance matrix to obtain eigenvalues ​​and corresponding eigenvectors.

[0159] Specifically, eigenvalue decomposition is performed on the covariance matrix to find the eigenvalues ​​and corresponding eigenvectors that satisfy the characteristic equation. The covariance matrix is ​​a second-order real symmetric matrix; eigenvalue decomposition yields two real eigenvalues ​​and their corresponding two mutually orthogonal eigenvectors. The eigenvector corresponding to the larger eigenvalue indicates the direction of maximum variance in the coordinate set data distribution, i.e., the main extension direction of the energy contour region; the eigenvector corresponding to the smaller eigenvalue indicates the direction of minimum variance in the data distribution, and is orthogonal to the main extension direction.

[0160] Step 702.4: Determine the eigenvector corresponding to the largest eigenvalue as the principal eigenvector, and obtain the crack direction information based on the angle between the principal eigenvector and the reference axis.

[0161] Specifically, the magnitudes of the two eigenvalues ​​obtained from eigenvalue decomposition are compared, and the eigenvector corresponding to the largest eigenvalue is determined as the principal eigenvector. The principal eigenvector indicates the main propagation direction of the region enclosed by the energy contour lines, which corresponds to the actual crack orientation. The angle between the principal eigenvector and a preset reference axis is calculated; the reference axis can be selected as the horizontal or vertical axis of the measured object's coordinate system. This angle is output as crack direction information, representing the crack's tilt direction relative to the reference axis in the form of an angle value, used for the complete reconstruction of crack features.

[0162] In one embodiment of this application, as Figure 8 As shown, in step 703, the energy profile is extracted along the direction corresponding to the crack direction information, and the crack size information is obtained based on the width range of the energy profile and the preset calibration relationship, including the following steps:

[0163] Step 703.1: Along the direction corresponding to the crack direction information, take a line in the energy field that passes through the precise positioning result to obtain a one-dimensional energy profile.

[0164] Specifically, the cross-sectional direction is determined based on the crack direction information obtained in step 702.4. Starting from the spatial coordinate point corresponding to the precise positioning result, a predetermined length is extended along the direction corresponding to the crack direction information and its opposite direction to form a cross-section passing through the precise positioning result. Energy values ​​at each location point are extracted along the cross-section in the energy field. The extracted energy values ​​are arranged in spatial order to obtain a one-dimensional energy profile. The one-dimensional energy profile, with spatial position as the horizontal axis and energy value as the vertical axis, characterizes the energy distribution characteristics of the energy field along the crack propagation direction. The width of the one-dimensional energy profile corresponds to the actual size of the crack.

[0165] Step 703.2: On the one-dimensional energy profile, determine two location points where the energy value drops to a preset percentage of the peak value, and calculate the distance between the two location points as the expansion range.

[0166] Specifically, the energy peak value is extracted from the one-dimensional energy profile, and the energy threshold is calculated based on the energy peak value and a preset ratio, which can be set to half, one-third, or other appropriate ratios of the peak value. On the one-dimensional energy profile, the points where the energy value first drops to the energy threshold are searched along the cutoff direction on both sides, obtaining two location points. The spatial distance between the two location points is calculated and used as the broadening range of the one-dimensional energy profile. The broadening range characterizes the spatial expansion of the time-reversed focused energy field in the crack propagation direction, and its magnitude is related to the actual length of the crack.

[0167] Step 703.3: Based on the calibration coefficients obtained in advance through the calibration of crack samples of known size, the width range is converted into crack size information.

[0168] Specifically, crack samples with different known sizes are pre-prepared. Guided wave detection and time-reversal imaging are performed on each crack sample to extract the one-dimensional energy profile spread range corresponding to each crack sample. A mapping relationship between the known size of each crack sample and its corresponding spread range is established. Calibration coefficients are obtained through linear or polynomial fitting. These calibration coefficients characterize the quantitative conversion relationship between the spread range and the actual crack size. The spread range obtained in step 801 is substituted into the conversion formula corresponding to the calibration coefficients to calculate the crack size information. The crack size information, together with the precise location results and crack direction information, constitute the complete feature reconstruction result of the crack, providing quantitative data support for the safety assessment and maintenance decisions of ship structures.

[0169] Figure 9 This is a schematic diagram of a ship structure crack location and feature reconstruction system provided in one embodiment of this application. Figure 9 As shown in the embodiment of this application, the ship structure crack location and feature reconstruction system includes: a sensor array 900, a signal acquisition module 901, a signal enhancement module 902, a feature extraction module 903, a preliminary positioning module 904, a precise positioning module 905, and a feature reconstruction module 906.

[0170] A sensor array 900 is arranged in the detection area of ​​the object under test to emit guided wave signals and acquire response signals. The sensor array 900 is connected to a signal acquisition module 901 to transmit the acquired response signals.

[0171] The signal acquisition module 901 is connected to the sensor array 900 and is used to record the position information of each sensor in the sensor array 900 and acquire the response signals collected by the sensor array 900 to form raw signal data. The signal acquisition module 901 is connected to the signal enhancement module 902 and transmits the raw signal data and position information to the signal enhancement module 902.

[0172] The signal enhancement module 902 is connected to the signal acquisition module 901 and is used to receive raw signal data. It performs Fourier transforms on the raw signal data along both the time and spatial dimensions to obtain the frequency-wavenumber domain signal distribution. The frequency-wavenumber domain signal distribution is compared and matched with the theoretical dispersion curve of the measured object to identify and filter target patterns. A bandpass frequency window is constructed based on the dispersion characteristics of each target pattern, and filtering is performed using the bandpass frequency window to obtain enhanced signal data. The signal enhancement module 902 is also connected to the feature extraction module 903 and the precise positioning module 905, transmitting the enhanced signal data to both modules.

[0173] The feature extraction module 903 is connected to the signal enhancement module 902 and is used to receive enhanced signal data, perform multi-mode feature extraction on the enhanced signal data, obtain the peak amplitude, time of arrival, group velocity change rate, phase difference and mode conversion energy ratio between each target mode, and construct a multi-dimensional feature vector containing the features of each target mode and the coupling features between modes. The feature extraction module 903 is also connected to the preliminary localization module 904 and transmits the multi-dimensional feature vector to the preliminary localization module 904.

[0174] The preliminary positioning module 904 is connected to the feature extraction module 903 and the signal acquisition module 901 respectively. It is used to receive multi-dimensional feature vectors and sensor position information. Based on the arrival time, corrected group velocity, peak amplitude, phase difference and mode conversion energy ratio in the multi-dimensional feature vectors and the sensor position information, the preliminary positioning result of the crack is obtained by establishing an elliptic equation system and solving it using a weighted optimization algorithm.

[0175] The precise positioning module 905 is connected to the signal enhancement module 902. It receives enhanced signal data, performs time-reversal processing and wavefield backpropagation calculations on the enhanced signal data, constructs an energy field through cross-correlation processing, and obtains the precise crack positioning result based on the extreme points of the energy field. This result is then compared and verified with the preliminary positioning result to confirm the final positioning result. The precise positioning module 905 is also connected to the feature reconstruction module 906, transmitting the energy field and precise positioning result to the feature reconstruction module 906.

[0176] The feature reconstruction module 906 is connected to the precise positioning module 905 and is used to receive the energy field and the precise positioning result. Based on the spatial distribution characteristics of the energy field around the precise positioning result, the direction and size information of the crack are obtained through principal component analysis and energy profile broadening calculation.

[0177] In another embodiment of the present application, the feature reconstruction module 906 in the ship structure crack location and feature reconstruction system includes: a contour extraction unit, a direction analysis unit, and a size calculation unit.

[0178] The contour line extraction unit is connected to the precise positioning module 905 and is used to receive the energy field and precise positioning results. Using the precise positioning results as the center, it extracts energy contour lines for the energy field within a local area. The contour line extraction unit is also connected to the direction analysis unit, transmitting the energy contour lines to the direction analysis unit.

[0179] The direction analysis unit is connected to the contour extraction unit. It receives energy contour lines, calculates the centroid of the set of coordinate points within the energy contour lines and performs centering, constructs the covariance matrix and performs eigenvalue decomposition, and obtains the crack direction information based on the direction of the principal eigenvector corresponding to the largest eigenvalue. The direction analysis unit is also connected to the size calculation unit, transmitting the crack direction information to the size calculation unit.

[0180] The size calculation unit is connected to the direction analysis unit and the precise positioning module 905 respectively. It is used to receive crack direction information, energy field and precise positioning results. It extracts a one-dimensional energy profile by taking the intercept line through the precise positioning result along the direction corresponding to the crack direction information. It determines two position points where the energy value drops to the preset percentage of the peak value. It calculates the distance between the two position points as the expansion range. It obtains crack size information based on the expansion range of the energy profile and the preset calibration coefficient.

[0181] It will be apparent to those skilled in the art that this application is not limited to the details of the exemplary embodiments described above, and that this application can be implemented in other specific forms without departing from the spirit or essential characteristics of this application. Therefore, the embodiments should be considered illustrative and non-limiting in all respects, and the scope of this application is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within this application. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. A method for locating and reconstructing features of cracks in ship structures, characterized in that, include: A sensor array is arranged in the area to be tested of the object being tested, and the position information of each sensor in the sensor array is recorded; The sensor array emits guided wave signals and acquires response signals to obtain raw signal data; The original signal data is subjected to domain transformation and screening of several target modes, and then filtered using a frequency window that matches each of the screened target modes to obtain enhanced signal data. Multi-mode feature extraction is performed on the enhanced signal data to construct a multi-dimensional feature vector containing the features of each target mode and the coupling features between modes; the features of each target mode include the peak amplitude, time of arrival, and group velocity change rate of each target mode, and the coupling features between modes include the phase difference and mode conversion energy ratio between each target mode; Based on the multidimensional feature vector and the position information of the sensor, the preliminary location result of the crack is obtained. The enhanced signal data is subjected to time reversal processing and wave field back propagation calculation to construct an energy field. The precise location result of the crack is obtained based on the extreme points of the energy field. The result is compared and verified with the preliminary location result to confirm the final location result. Based on the spatial distribution characteristics of the energy field around the precise positioning result, the direction and size information of the crack are obtained, and feature reconstruction is completed.

2. The method for locating and reconstructing ship structural cracks according to claim 1, characterized in that, The process of performing domain transformation and target mode filtering on the original signal data, and then filtering it using a frequency window matching the selected target mode to obtain enhanced signal data, includes: The original signal data is subjected to Fourier transform along the time and spatial dimensions to obtain the frequency-wavenumber domain signal distribution; The frequency-wavenumber domain signal distribution is compared and matched with the theoretical dispersion curve of the object under test to identify each guided wave mode and its energy distribution. Based on the displacement field distribution characteristics, scattering energy ratio and mode separation degree of each guided wave mode, the target mode is selected. Based on the dispersion characteristics of each target mode, the concentrated frequency range corresponding to each target mode is determined, and a bandpass frequency window is constructed within each concentrated frequency range; the concentrated frequency range is specifically the range in which the energy corresponding to the current target mode accounts for a proportion of the total energy of the entire frequency band greater than 0.7, and the ratio of the distance between the wavenumber of the current target mode and the wavenumber of the adjacent mode is greater than 0.

15. The original signal data is filtered using each of the aforementioned bandpass frequency windows to obtain the signal components corresponding to each target mode. The signal components corresponding to each target mode constitute the enhanced signal data.

3. The method for locating and reconstructing ship structural cracks according to claim 2, characterized in that, The process of selecting target modes based on the displacement field distribution characteristics, scattering energy ratio, and mode separation of each guided wave mode includes: In the frequency-wavenumber domain signal distribution, the ratio of the energy of each guided wave mode to the energy of the direct wave within the time window corresponding to the crack reflection wave is calculated, and the ratio is defined as the scattered energy ratio. Guided wave modes with a scattered energy ratio higher than a preset energy ratio threshold are selected as candidate target modes. In the frequency-wavenumber domain, the mode separation degree between each candidate target mode and other modes is evaluated. The mode separation degree is defined as the ratio of the wavenumber difference between adjacent modes at the same frequency to the average wavenumber. Modes with a mode separation degree greater than a preset separation threshold are selected as target modes.

4. The method for locating and reconstructing ship structural cracks according to claim 1, characterized in that, The step of performing multi-mode feature extraction on the enhanced signal data to construct a multi-dimensional feature vector containing features of each target mode and coupling features between modes includes: Envelope extraction is performed on the signal components of each target mode in the enhanced signal data to obtain the peak amplitude and arrival time of each target mode; The measured group velocity of each target mode is calculated based on the distance between signal acquisition points and the arrival time of the direct wave. The measured group velocity is then compared with the theoretical group velocity to obtain the rate of change of the group velocity of each target mode. Instantaneous phase extraction is performed on the signal components of each target mode, and the phase difference between each target mode at the crack reflection wave is calculated. Calculate the ratio of mode conversion energy to total reflected energy among the target modes to obtain the mode conversion energy ratio; The peak amplitude, arrival time, group velocity change rate, phase difference, and mode transition energy ratio of each target mode are integrated to construct the multidimensional feature vector.

5. The method for locating and reconstructing ship structural cracks according to claim 1, characterized in that, The step of obtaining the preliminary crack location result based on the multidimensional feature vector and the sensor's position information includes: The theoretical group velocity of each target mode is corrected based on the group velocity change rate in the multidimensional feature vector to obtain the corrected group velocity. Based on the arrival time of each target mode in the multidimensional feature vector and the corrected group velocity, the propagation path length of the guided wave from the sensor's signal emission point to the receiving point via crack reflection is calculated. Based on the propagation path length and the position information of each sensor, an elliptical equation is established with the crack location coordinates as unknowns; Based on the phase difference between each target pattern in the multidimensional feature vector, data whose phase difference exceeds a preset phase difference threshold are removed. Based on the mode conversion energy ratio in the multidimensional feature vector, mode combinations with a mode conversion energy ratio higher than a preset energy ratio threshold are selected to participate in the localization calculation. An overdetermined system of equations is established using multiple sets of propagation path lengths. Weighting coefficients are constructed based on the peak amplitude of each target mode in the multidimensional feature vector. The weighting coefficients are obtained by normalizing the peak amplitude of each propagation path by dividing it by the maximum value of the peak amplitude of all propagation paths. The values ​​range from 0 to 1. The overdetermined system of equations is solved using a weighted optimization algorithm to obtain the preliminary location results of the crack.

6. The method for locating and reconstructing cracks in ship structures according to claim 1, characterized in that, The enhanced signal data undergoes time-reversal processing and wavefield backpropagation calculation to construct an energy field. Based on the extreme points of this energy field, the precise location of the crack is obtained. This is then compared and verified with the preliminary location results to confirm the final location. This process includes: The sampled value sequences of each signal channel in the enhanced signal data are arranged in reverse order along the time axis to obtain the time-reversed signal; The time-reversed signal is used as a boundary condition input to the wave equation numerical model to calculate the backward propagation process of the guided wave in the measured object and obtain the displacement field distribution at each time. Cross-correlation processing is performed on the backpropagation sound fields of each sensor channel, and the product of the sound fields of different channels is integrated along the time axis to construct a cross-energy field as the energy field. The coordinates of the point with the maximum energy value are extracted from the energy field to obtain the precise location of the crack. The precise positioning result is compared and verified with the preliminary positioning result. If the deviation between the two is within a preset threshold range, the precise positioning result is confirmed as the final positioning result.

7. The method for locating and reconstructing cracks in ship structures according to claim 1, characterized in that, The step of obtaining crack direction and size information based on the spatial distribution characteristics of the energy field around the precise positioning result includes: Using the precise positioning result as the center, extract the energy contour lines of the energy field within a local area of ​​a preset range; Principal component analysis is performed on the set of coordinate points within the energy contour lines to obtain crack direction information based on the direction of the principal eigenvectors. Energy profiles are extracted along the direction corresponding to the crack direction information, and crack size information is obtained based on the width range of the energy profiles and a preset calibration relationship.

8. The method for locating and reconstructing ship structural cracks according to claim 7, characterized in that, The step of performing principal component analysis on the set of coordinate points within the energy contour lines, obtaining crack direction information based on the direction of the principal eigenvectors, and completing feature reconstruction includes: Extract the coordinates of all points within the energy contour lines and calculate the centroid of the coordinate set; The coordinate set is decentered using the centroid as the origin to construct a covariance matrix; The covariance matrix is ​​decomposed into eigenvalues ​​to obtain eigenvalues ​​and corresponding eigenvectors. The eigenvector corresponding to the largest eigenvalue is determined as the main eigenvector, and the crack direction information is obtained based on the angle between the main eigenvector and the reference axis.

9. The method for locating and reconstructing ship structural cracks according to claim 7, characterized in that, The step of extracting an energy profile along the direction corresponding to the crack direction information, and obtaining crack size information based on the width range of the energy profile and a preset calibration relationship, includes: Along the direction corresponding to the crack direction information, take a line through the precise positioning result in the energy field to obtain a one-dimensional energy profile; On the one-dimensional energy profile, two location points are determined where the energy value drops to a preset percentage of the peak value, and the distance between the two location points is calculated as the expansion range. The width expansion range is converted into crack size information based on the calibration coefficients obtained in advance through calibration of crack samples of known size.

10. A system for locating and reconstructing cracks in ship structures, characterized in that, include: A sensor array is arranged in the area to be detected of the object being measured, and is used to emit guided wave signals and acquire response signals; The signal acquisition module is used to record the position information of each sensor in the sensor array and to acquire the response signals collected by the sensor array to form raw signal data; The signal enhancement module is used to perform Fourier transform on the original signal data along the time and space dimensions to obtain the frequency-wavenumber domain signal distribution, compare and match the frequency-wavenumber domain signal distribution with the theoretical dispersion curve of the object under test to identify and filter target modes, construct a bandpass frequency window according to the dispersion characteristics of each target mode, and use the bandpass frequency window for filtering to obtain enhanced signal data. The feature extraction module is used to perform multi-mode feature extraction on the enhanced signal data, obtain the peak amplitude, arrival time, group velocity change rate of each target mode, as well as the phase difference and mode conversion energy ratio between each target mode, and construct a multi-dimensional feature vector containing the features of each target mode and the coupling features between modes. The preliminary positioning module is used to obtain the preliminary positioning result of the crack by establishing an elliptic equation system and solving it using a weighted optimization algorithm based on the arrival time, corrected group velocity, peak amplitude, phase difference, and mode conversion energy ratio in the multidimensional feature vector and the position information of the sensor. The precise positioning module is used to perform time reversal processing and wave field back propagation calculation on the enhanced signal data, construct an energy field through cross-correlation processing, obtain the precise positioning result of the crack based on the extreme points of the energy field, compare and verify with the preliminary positioning result, and confirm the final positioning result. The feature reconstruction module is used to obtain the direction and size information of the crack through principal component analysis and energy profile broadening calculation based on the spatial distribution characteristics of the energy field around the precise positioning result, and to complete the feature reconstruction.

Images