Hidden damage diagnosis method of longitudinal joints in shield tunnels based on multimodal ultrasonic guided waves

Abstract

The present invention provides a method for diagnosing hidden damage in longitudinal seam joints of shield tunnels based on multimodal ultrasonic guided waves. The method includes applying orthogonal polarization in specific areas of the longitudinal seam joint to form a time-space coupled guided wave interference pattern over the entire longitudinal seam joint; utilizing the obtained guided wave interference pattern to separate the baseline interference characteristics and the abnormal disturbance component caused by hidden damage through multi-physical field reverse decoupling, wherein the abnormal disturbance component manifests as abnormal energy attenuation and phase distortion in a specific polarization direction; constructing a three-dimensional damage sensitive vector field inside the longitudinal seam joint; utilizing the generated three-dimensional damage sensitive vector field to perform dynamic manifold topological analysis, identify topological features in the three-dimensional damage sensitive vector field, and track the migration path and intensity evolution trajectory of the topological features as the detection conditions change. The present invention realizes full-dimensional diagnosis and evolution prediction of hidden damage in longitudinal seam joints of shield tunnels.

Description

Technical Field

[0001] The present invention relates to the technical field of longitudinal seam joint damage judgment, and in particular to a method for diagnosing hidden damage of longitudinal seam joints in a shield tunnel based on multi-modal ultrasonic guided waves. Background Art

[0002] The expansion of modern transportation networks has driven a surge in tunnel construction. Longitudinal joints, the core connecting components of precast concrete segments, face a significant risk of hidden damage. The industry currently faces three challenges: Concealment: Damage within longitudinal joints, such as corrosion of rebar and microcracks in concrete, is concealed by a waterproof sealant, leading to a 67% failure rate for traditional visual inspections. There is also a time lag: an average incubation period of 8-12 months from damage initiation to visible cracks. Furthermore, there is a cascading risk: damage to a single longitudinal joint can trigger stress redistribution in adjacent segments, leading to a domino-like pattern of structural degradation.

[0003] Prior art 1, application number: 202311008097.6 discloses a method and device for ultrasonically detecting the health status of longitudinal seam joints in an operating shield tunnel, the method comprising: starting a concrete ultrasonic detector and recording the sound time of ultrasonic waves propagating from a transmitting probe to a receiving probe; the transmitting probe remains stationary, and the receiving probe is moved an equal distance away from the longitudinal seam each time, starting the concrete ultrasonic detector, and recording the sound time of ultrasonic waves propagating from the transmitting probe to the receiving probe at each distance in turn; the transmitting probe and the receiving probe are exchanged, the transmitting probe is fixed on the side of the original receiving probe near the edge of the longitudinal seam and remains stationary, and the receiving probe is on the side of the original transmitting probe, and the sound time at different distances is repeatedly measured. The distance between the receiving probe and the edge of the longitudinal seam when the sound time is minimum is the length of the fracturing crack, and the degree of damage to the longitudinal seam joint is judged by the length of the fracturing crack; the device comprises: an ultrasonic detector, a transmitting probe, a receiving probe and a longitudinal seam. Although it can provide effective data support for health assessments of currently operating shield tunnels, it only relies on acoustic-time measurement: using single-probe ultrasonic detection, only the ultrasonic propagation time is recorded, and the degree of damage is indirectly judged by the length of the fracturing crack, which cannot accurately identify hidden damage; the detection dimension is single: only one-dimensional assessment is based on acoustic-time changes, which cannot reflect the spatial distribution of damage, modal conversion characteristics and damage evolution trends; the anti-interference ability is weak: it is greatly affected by factors such as the anisotropy of concrete materials and environmental noise, and it is difficult to distinguish between geometric discontinuities and real damage; it cannot achieve three-dimensional damage characterization: it can only provide simple parameters such as crack length, and cannot construct a vector field or topological evolution model of damage.

[0004] Prior art 2, application number 201010289974.8, discloses a multifunctional lining segment joint mechanical properties testing system capable of three-dimensional loading. The system primarily comprises a self-balancing frame subsystem, a loading subsystem, a specimen support subsystem, a specimen transport subsystem, a loading control subsystem, and a data acquisition subsystem. The loading subsystem utilizes a modular design, and through different combinations, it can perform tests such as inter-annular shear testing, moment transfer testing, longitudinal joint angular stiffness testing, and longitudinal joint radial shear testing on shield tunnel lining segments. While it effectively simulates the three stress states experienced by shield tunnel lining segment joints in real-world conditions and can handle a variety of loading methods to measure the mechanical parameters of shield tunnel lining segment joints and the mechanical properties of other similar structures, it is only suitable for laboratory loading tests: it relies on mechanical loading to simulate stress states and cannot be directly used for on-site nondestructive testing. It cannot identify hidden damage: it can only measure macroscopic mechanical parameters and cannot detect early-stage damage such as microcracks and material degradation. Furthermore, it lacks dynamic monitoring: it is a static or quasi-static test and cannot track damage propagation in real time.

[0005] Currently, existing technologies 1 and 2 have problems such as inability to accurately identify hidden damage, lack of three-dimensional dynamic damage characterization capabilities, and poor anti-interference performance. Therefore, the present invention provides a method for diagnosing hidden damage in longitudinal seam joints of shield tunnels based on multimodal ultrasonic guided waves. Summary of the Invention

[0006] In order to solve the above technical problems, the present invention provides a method for diagnosing hidden damage of longitudinal seam joints in shield tunnels based on multimodal ultrasonic guided waves, which includes the following steps: using the generated three-dimensional damage-sensitive vector field to perform dynamic manifold topology analysis, identify the topological characteristics of the energy flow convergence points, divergence points of the damage extension front, and vector curl anomaly areas of the material performance degradation zone in the three-dimensional damage-sensitive vector field, and track the migration path and intensity evolution trajectory of the topological characteristics as the detection conditions change; determine the dominant mode of the hidden damage and the critical state level of its development based on the migration path stability and intensity evolution rate of the topological characteristics.

[0007] Optionally, the process of determining the dominant mode of latent damage and the criticality level of its development includes the following steps:

[0008] The generated three-dimensional damage-sensitive vector field is input and the vector field time series data of the same joint area is obtained under the time-varying detection conditions of different preset mechanical loads and temperature and humidity states to form a dynamic vector field sequence. Each time slice contains the full space vector distribution.

[0009] The local manifold structure of the vector field of each time slice in the three-dimensional damage-sensitive vector field is analyzed. The positions of three types of features, namely energy flow convergence points, divergence points, and vector curl anomaly areas, and the corresponding vector modulus mean and directional attributes are integrated into a spatiotemporal topological feature set. The state matrix of the time-varying topological features is formed. The rows of the state matrix record different time points, and the columns record the spatial coordinates, intensity, direction angle, and regional size of the three types of features.

[0010] The spatial coordinate sequence of each type of feature in the state matrix is ​​subjected to a convergence analysis of the intensity evolution trajectory, and the standard deviation of the angles between the displacement vectors at adjacent time points is calculated. The intensity sequence of each type of feature in the state matrix is ​​subjected to a time-domain gradient integration to obtain the intensity change per unit time. This forms a multi-parameter evolution state diagnosis of latent damage, including the dominant damage mode and critical level.

[0011] Optionally, the process of forming a state matrix of time-varying topological features includes the following steps:

[0012] The generated dynamic vector field sequence is input and in each time slice, the entire joint domain is divided into overlapping local manifold cells. The cell size is constrained by the joint geometry characteristics to ensure that the minimum damage scale is covered.

[0013] Energy flow convergence point detection, divergence point detection, and vector curl anomaly area detection are performed on each local manifold unit; the spatial position, intensity index, and dominant direction of the three types of features are bound into a feature tuple to form a time-space-strong coupling original feature set;

[0014] All features in the original time-space-strong coupling feature set are classified and aggregated according to timestamps and feature types; multiple instances of each type of feature at the same time are spatially clustered and merged; for the merged features, their spatial centroid coordinates, the mean of the merged strength index and the composite value of the dominant direction are calculated; the merged features at each time point are filled with matrix rows according to a fixed dimension to form a state matrix of time-varying topological features.

[0015] Optionally, the process of forming a multi-parameter evolution state diagnosis of hidden damage includes the following steps:

[0016] The generated state matrix is ​​input, and its row vectors contain the spatial coordinate sequence, intensity sequence, and direction angle sequence of three types of features: energy flow convergence point, divergence point, and vector curl anomaly area in time series;

[0017] For each type of feature's spatial coordinate sequence, the coordinate differences between adjacent time points are calculated to form a displacement vector sequence. The standard deviation of the directional angles of all adjacent displacement vectors in the spatial coordinate sequence is then calculated to generate a path fluctuation coefficient. For each type of feature's intensity sequence, the intensity differences between adjacent time points are calculated, and the absolute value of the difference sequence is accumulated along the time axis to generate an intensity transition cumulant. This forms a dual-channel parameter set for coupled spatial and intensity evolution, with each type of feature corresponding to a path fluctuation coefficient and an intensity transition cumulant.

[0018] If the path fluctuation coefficient of the divergence point is greater than the first preset threshold and its strength transition cumulative amount is the largest, it is determined to be dominated by tensile debonding; if the strength transition cumulative amount of the vector curl anomaly area and the path fluctuation coefficient of the energy flow convergence point increase synchronously, it is determined to be dominated by material degradation; the critical level is determined based on the relationship between the strength transition cumulative amount and the preset threshold.

[0019] Optional, critical state mapping:

[0020] When tensile debonding is dominant, the cumulative amount of strength transition is less than the second preset threshold, which is a stable latent period; when the cumulative amount of strength transition is between the second preset threshold and the third preset threshold, including the second preset threshold and the third preset threshold, it is a subcritical expansion period; when the cumulative amount of strength transition is greater than the third preset threshold, it is close to the instability period;

[0021] When material degradation is dominant: the path fluctuation coefficient of the energy flow convergence point is less than the fourth preset threshold value, which is a stable latent period; the ratio of the curl zone to the convergence point parameter is greater than the fifth preset threshold value, which is close to the instability period.

[0022] Optionally, the process of generating the path fluctuation coefficient and the intensity transition cumulative amount includes the following steps:

[0023] The input state matrix row vector is used to extract the spatial coordinate time series chain of the three-dimensional coordinates of each type of characteristic continuous time point and the intensity value time series chain of the intensity index value of the continuous time point; thus forming the original sequence of the characteristic spatiotemporal evolution;

[0024] For the spatial coordinate time series chain of each feature in the original sequence of the spatiotemporal evolution of the features, the vector difference of the coordinates of adjacent time points is calculated to generate a displacement vector sequence; for the intensity value time series chain of each feature in the original sequence of the spatiotemporal evolution of the features, the algebraic difference of the intensity values ​​of adjacent time points is calculated to generate an intensity difference sequence; a dual-channel evolution field of coupled spatial motion and intensity change is formed, in which the displacement vector sequence describes the evolution path of the feature position, and the intensity difference sequence describes the transition behavior of the feature intensity;

[0025] The directional unit vectors of all adjacent displacement vectors are extracted from the displacement vector sequence of the dual-channel evolution field, the dot product arc cosine value of each pair of adjacent directional vectors is calculated, the standard deviation of the dot product arc cosine value is obtained, and the path fluctuation coefficient is output; the absolute value of the intensity difference sequence of the dual-channel evolution field is taken, the absolute value sequence is linearly weighted accumulated along the time axis, and the intensity transition accumulation is output; a dual-channel evolution parameter set is formed, and each feature corresponds to two parameters: the path fluctuation coefficient and the intensity transition accumulation.

[0026] Optionally, each vector in the displacement vector sequence represents the spatial migration direction and distance of the feature in unit time; a positive value in the intensity difference sequence indicates enhancement, and a negative value indicates weakening.

[0027] Optionally, the process of generating a three-dimensional damage-sensitive vector field includes the following steps: using the acquired guided wave interference pattern, through multi-physical field reverse decoupling, separating the baseline interference characteristics caused by the geometric discontinuity of the joint in the guided wave interference pattern, and the abnormal disturbance component caused by the hidden damage, the abnormal disturbance component manifests as abnormal energy attenuation and phase distortion in a specific polarization direction; forming the spatial distribution and intensity of the abnormal disturbance component based on the multi-physical field reverse decoupling, constructing a three-dimensional damage-sensitive vector field inside the longitudinal seam joint, and quantifying the response intensity and directionality of different positions to the damage pattern.

[0028] Optionally, the process of constructing a three-dimensional damage sensitivity vector field inside the longitudinal seam joint includes the following steps:

[0029] The abnormal disturbance component output by the reverse decoupling of the multi-physics field is discretely sampled in the polarization direction according to a preset spatial grid in the entire longitudinal seam joint. Each sampling point captures the disturbance intensity and phase offset angle in a specific polarization plane.

[0030] Perform a three-dimensional spatial gradient calculation on the energy attenuation intensity of each sampling point in the abnormal disturbance component to generate an energy attenuation gradient vector field; associate the phase distortion value of each sampling point in the abnormal disturbance component with the preset polarization direction to construct a polarization-phase correlation matrix; and form a dual-channel spatial field with coupled gradient and polarization properties.

[0031] The direction of the energy attenuation gradient vector field in the dual-channel spatial field is taken as the direction of the basis vector. The modulus of the basis vector is adjusted according to the weight of the polarization-phase correlation matrix in the dual-channel spatial field. The second-order derivative of the phase distortion space is introduced to correct the direction of the basis vector. The corrected basis vector is subjected to directional enhancement filtering to retain the potential damage extension direction of the seam, forming a three-dimensional damage-sensitive vector field.

[0032] Optionally, the process of generating the guided wave interference pattern includes the following steps: applying orthogonal polarization in a specific area of ​​the longitudinal seam joint, a composite ultrasonic energy field containing longitudinal and shear wave energy flows of a specific frequency combination; forming a time-space coupled guided wave interference pattern in the entire domain of the longitudinal seam joint, including the joint body and the concrete on both sides, the guided wave interference pattern including the amplitude distribution, phase delay and modal conversion characteristics under the interaction of the composite ultrasonic energy field.

[0033] The present invention uses a composite ultrasonic energy field to excite multimodal guided wave interference across the entire joint, coupling the amplitude-phase-mode conversion characteristics, breaking through the dimensional limitations of traditional single-mode detection and achieving a three-dimensional sensitive response to hidden damage. Multi-physics field reverse decoupling technology extracts damage characteristic components from complex interference patterns, and the constructed three-dimensional vector field can simultaneously analyze the spatial distribution pattern of damage, the directionality of mechanical response, and the material degradation trend. Based on the manifold topological analysis of the vector field, a mapping relationship between damage characteristics and material performance degradation is established. The damage core is identified by the energy flow convergence point, the expansion front is determined by the divergence point, and the performance degradation zone is characterized by the curl anomaly area, forming a complete topological description system for damage evolution. The migration path tracking technology reveals the dynamic laws of damage expansion, achieving a leap from static detection to dynamic prediction. Orthogonal polarization excitation enhances sensitivity to microscopic defects, vector field construction realizes mesoscopic damage pattern recognition, and topological analysis completes the prediction of macroscopic evolution trends, forming a cross-scale diagnostic chain. It effectively distinguishes the baseline characteristics caused by the geometric discontinuity of the joint from the actual damage disturbance, significantly improving the detection ability of early hidden damage. The time-space coupled guided wave interference mechanism suppresses material anisotropic interference through modal conversion characteristics, and the multi-physics field decoupling algorithm automatically compensates for environmental noise, enabling the system to maintain stable diagnostic accuracy under complex working conditions; the entire technical system realizes closed-loop diagnosis from damage location, pattern recognition to status assessment.

[0034] Other features and advantages of the present invention will be described in the following description, and in part will become apparent from the description, or will be understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description and the accompanying drawings.

[0035] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0037] Figure 1 This is a flow chart of the method for diagnosing hidden damage of longitudinal seam joints in a shield tunnel based on multimodal ultrasonic guided waves in Example 1 of the present invention;

[0038] Figure 2 This is a schematic diagram of the hidden damage diagnosis method for longitudinal seam joints in shield tunnels based on multi-modal ultrasonic guided waves in Example 1 of the present invention;

[0039] Figure 3 This is a diagram showing the process of forming a spatiotemporally coupled guided wave interference pattern in Example 2 of the present invention;

[0040] Figure 4 This is a diagram showing the process of constructing a three-dimensional damage-sensitive vector field inside a longitudinal seam joint in Example 3 of the present invention;

[0041] Figure 5 This is a process diagram for determining the dominant mode of latent damage and the critical state level of its development in Example 5 of the present invention. DETAILED DESCRIPTION

[0042] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0043] The terms used in the embodiments of the present application are for the purpose of describing specific embodiments only and are not intended to limit the embodiments of the present application. The singular forms "a", "the" and "the" used in the embodiments of the present application are also intended to include plural forms unless the context clearly indicates otherwise. It should also be understood that the term "and / or" used herein refers to and includes any or all possible combinations of one or more associated listed items.

[0044] When the following description refers to the accompanying drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present application. On the contrary, they are merely examples of devices and methods consistent with some aspects of the present application. In the description of the present application, it should be understood that the terms "first", "second", "third", etc. are only used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence, nor can they be understood as indicating or implying relative importance. For those of ordinary skill in the art, the specific meanings of the above terms in the present application can be understood according to the specific circumstances.

[0045] Example 1: Figure 1 As shown, an embodiment of the present invention provides a method for diagnosing hidden damage of longitudinal seam joints in a shield tunnel based on multimodal ultrasonic guided waves, comprising the following steps:

[0046] S100: A composite ultrasonic energy field with orthogonal polarization, including longitudinal and shear wave energy flows of a specific frequency combination, is applied to a specific area of ​​the longitudinal joint. A spatiotemporally coupled guided wave interferometry pattern is formed over the entire longitudinal joint, including the joint body and the concrete on both sides. The guided wave interferometry pattern includes the amplitude distribution, phase delay, and modal conversion characteristics under the interaction of the composite ultrasonic energy field.

[0047] S200: Using the acquired guided wave interferometer pattern, through multi-physics inverse decoupling, the baseline interference characteristics caused by the joint geometric discontinuity and the abnormal disturbance components caused by hidden damage in the guided wave interferometer pattern are separated. The abnormal disturbance components manifest as abnormal energy attenuation and phase distortion in a specific polarization direction. The spatial distribution and intensity of the abnormal disturbance components derived from multi-physics inverse decoupling are formed, and a three-dimensional damage-sensitive vector field inside the longitudinal seam joint is constructed to quantify the response intensity and directionality of different locations to the damage pattern.

[0048] S300: Use the generated three-dimensional damage-sensitive vector field to perform dynamic manifold topological analysis to identify the topological characteristics of the three-dimensional damage-sensitive vector field, including the energy flow convergence points of potential damage cores, the divergence points of the damage extension front, and the vector curl anomaly areas of the material performance degradation zone. Track the migration path and intensity evolution trajectory of the topological characteristics as the detection conditions change; determine the dominant mode of hidden damage and the critical state level of its development based on the stability of the migration path and the intensity evolution rate of the topological characteristics.

[0049] The working principle and beneficial effects of the above technical solution are as follows: In this embodiment, a composite ultrasonic energy field with orthogonal polarization, including longitudinal and shear wave energy flows of a specific frequency combination, is first applied to a specific area of ​​the longitudinal seam joint; a spatiotemporally coupled guided wave interference pattern is formed in the entire longitudinal seam joint, including the joint body and the concrete on both sides. The guided wave interference pattern includes the amplitude distribution, phase delay and modal conversion characteristics under the interaction of the composite ultrasonic energy field; secondly, the obtained guided wave interference pattern is used to separate the baseline interference characteristics caused by the geometric discontinuity of the joint in the guided wave interference pattern through multi-physical field reverse decoupling. The longitudinal seam joint is a continuous structural segment formed by adjacent pipe segments connected by bolts, sealing materials, etc., which undertakes comprehensive functions such as load transfer and waterproofing; the joint body specifically refers to the contact interface between the concrete pipe segments inside the joint (including filling materials, embedded parts, etc.), which is the main area of ​​stress concentration and damage initiation). As well as the abnormal disturbance component caused by latent damage, the abnormal disturbance component manifests as abnormal energy attenuation and phase distortion in a specific polarization direction; the spatial distribution and intensity of the abnormal disturbance component based on the inverse decoupling of multi-physical fields are formed, and a three-dimensional damage sensitive vector field inside the longitudinal seam joint is constructed to quantify the response intensity and directionality of different positions to the damage mode; finally, the generated three-dimensional damage sensitive vector field is used to perform dynamic manifold topological analysis to identify the topological features of the energy flow convergence point containing the potential damage core, the divergence point of the damage extension front, and the vector curl abnormal area of ​​the material performance degradation zone in the three-dimensional damage sensitive vector field, and track the migration path and intensity evolution trajectory of the topological features as the detection conditions change; according to the migration path stability and intensity evolution rate of the topological features, the dominant mode of the latent damage and the critical state level of its development are determined (the principle is referred to the attached Figure 2). The above scheme realizes the full-dimensional diagnosis and evolution prediction of hidden damage in the longitudinal joints of shield tunnels through the synergistic effect of multi-modal excitation of orthogonal polarized composite ultrasonic energy fields, multi-physics field inverse decoupling and dynamic manifold topological analysis. Its comprehensive technical effect is manifested as follows: through the multi-modal guided wave interference excited by the composite ultrasonic energy field in the entire joint area, the amplitude-phase-mode conversion characteristics are coupled, breaking through the dimensional limitation of traditional single-mode detection, and realizing a three-dimensional sensitive response to hidden damage; the multi-physics field inverse decoupling technology extracts the damage characteristic components from the complex interference pattern, and the constructed three-dimensional vector field can simultaneously analyze the spatial distribution pattern of the damage, the directionality of the mechanical response and the material degradation trend. Based on the manifold topological analysis of the vector field, a mapping relationship between the damage characteristics and the degradation of material performance is established. The damage core is identified by the energy flow convergence point, the expansion front is determined by the divergence point, and the performance degradation zone is characterized by the curl anomaly area, forming a complete topological description system for damage evolution; the migration path tracking technology reveals the dynamic law of damage expansion, realizing the leap from static detection to dynamic prediction. Orthogonal polarization excitation enhances sensitivity to microscopic defects, vector field construction enables mesoscopic damage pattern recognition, and topological analysis predicts macroscopic evolution trends, forming a cross-scale diagnostic chain. This effectively distinguishes baseline characteristics caused by geometric discontinuities in joints from actual damage disturbances, significantly improving the ability to detect early-stage hidden damage. A spatiotemporally coupled guided wave interferometry mechanism suppresses material anisotropic interference through modal conversion characteristics, and a multi-physics field decoupling algorithm automatically compensates for ambient noise, enabling the system to maintain stable diagnostic accuracy even under complex operating conditions. The entire technical system achieves closed-loop diagnostics from damage location, pattern recognition, and condition assessment.

[0050] Example 2: Figure 3 As shown, based on Example 1, the process of forming a spatiotemporally coupled guided wave interferogram provided by the embodiment of the present invention includes the following steps:

[0051] S101: Applying orthogonally polarized composite ultrasonic energy fields to specific areas of the longitudinal seam to penetrate the seam interface in a non-contact manner;

[0052] S102: When the longitudinal wave energy flow of the specific frequency combination of the composite ultrasonic energy field propagates along the length direction of the joint, it repeatedly interacts with the joint geometric interface to generate a periodic phase-modulated wave train. The shear wave energy flow of the composite ultrasonic energy field forms a polarization-constrained shear waveguide in the concrete on both sides of the joint. The direction of the energy flow is dynamically deflected by the anisotropic properties of the material, forming a coupled wave group dynamic propagation network in the entire longitudinal joint, including real-time energy exchange between the longitudinal wave phase-modulated wave train and the shear wave polarization-constrained waveguide.

[0053] S103: The coupled wave group dynamic propagation network is triggered at the interface between the joint body and the concrete. The longitudinal wave phase modulation wave train and the transverse wave polarization constrained waveguide energy flow vector are superimposed, resulting in a spatially non-uniform distribution of extreme points of interference energy density. The transverse wave polarization plane is scattered by the concrete microstructure, causing the accumulation of phase lag related to the polarization direction. The longitudinal and transverse wave energy flows undergo asymmetric modal conversion in the damaged area, forming a time-space coupled guided wave interference pattern.

[0054] The working principle and beneficial effects of the above technical solution are as follows: First, in this embodiment, a composite ultrasonic energy field with orthogonal polarization is applied to a specific area of ​​the longitudinal joint street to penetrate the joint interface in a non-contact manner; second, when the longitudinal wave energy flow of the specific frequency combination of the composite ultrasonic energy field propagates along the length direction of the joint, it repeatedly interacts with the geometric interface of the joint to generate a periodic phase-modulated wave train; the shear wave energy flow of the composite ultrasonic energy field forms a polarization-constrained shear waveguide in the concrete on both sides of the joint. The direction of its energy flow is dynamically deflected by the anisotropy of the material, forming a coupled wave group dynamic propagation network in the entire longitudinal joint, including real-time energy exchange between the longitudinal wave phase-modulated wavetrain and the shear wave polarization-constrained waveguide; finally, the coupled wave group dynamic propagation network is triggered at the junction of the joint body and the concrete, and the energy flow vectors of the longitudinal wave phase-modulated wavetrain and the shear wave polarization-constrained waveguide are superimposed, resulting in a spatially non-uniform distribution of extreme points of interference energy density; the shear wave polarization plane is scattered by the concrete microstructure, causing accumulation of phase lag related to the polarization direction; the longitudinal and shear wave energy flows undergo asymmetric modal conversion in the damaged area, forming a time-space coupled guided wave interference pattern. The above scheme constructs a dynamic wave group propagation network within the longitudinal joint region through the multimodal coupling of orthogonally polarized composite ultrasonic energy fields, achieving precise spatiotemporal control of the energy distribution within the structure. The orthogonally polarized longitudinal / transverse composite energy fields penetrate the interface through contactless excitation. The longitudinal wave energy flow forms a phase-modulated wave train along the length of the joint, while the transverse wave energy flow establishes a polarization-constrained waveguide within the concrete. The two achieve dynamic energy exchange through material anisotropy, forming a coupled wave group network covering the entire joint. The wave group network triggers the vector superposition of longitudinal and transverse wave energy flows at the joint interface, generating a spatially inhomogeneous distribution of interference extreme points. The shear wave polarization plane is scattered by the microstructure, resulting in phase lag accumulation. Asymmetric modal conversion in the damaged region further enhances the spatiotemporal coupling characteristics of the interference pattern. Through real-time energy exchange between the phase-modulated wave train and the polarization-constrained waveguide, the coupled wave group network maps joint geometry, material anisotropy, and damage state into spatiotemporal modulation signals within the interference pattern, enabling sensitive cross-scale characterization of internal structural defects.

[0055] In summary, this embodiment forms a guided wave interference pattern with time and space resolution capabilities, providing a composite diagnostic basis for the detection of hidden defects in joint structures by coupling the dynamic propagation field with the static interference field.

[0056] Example 3: Figure 4As shown, based on Example 1, the process of constructing a three-dimensional damage-sensitive vector field inside a longitudinal seam joint provided by the embodiment of the present invention includes the following steps:

[0057] S201: The abnormal disturbance component output by the inverse decoupling of the multi-physical field is discretely sampled in the polarization direction according to a preset spatial grid in the entire longitudinal seam joint. Each sampling point captures the disturbance intensity and phase offset angle in a specific polarization plane.

[0058] S202: performing a three-dimensional spatial gradient operation on the energy attenuation intensity of each sampling point in the abnormal disturbance component to generate an energy attenuation gradient vector field, the direction of which points to the direction of the maximum rate of change of energy attenuation, and the modulus represents the rate of change; correlating the phase distortion value of each sampling point in the abnormal disturbance component with a preset polarization direction to construct a polarization-phase correlation matrix, wherein the matrix elements quantify the contribution weight of a specific polarization direction to the phase distortion; forming a dual-channel spatial field of coupled gradient and polarization properties, including the energy attenuation gradient vector field and the polarization-phase correlation matrix;

[0059] S203: The direction of the energy attenuation gradient vector field in the dual-channel spatial field is used as the basic vector direction. The polarization-phase correlation matrix in the dual-channel spatial field is used to adjust the modulus of the basic vector according to the weight. The second-order derivative of the phase distortion space is introduced to correct the basic vector direction. The second-order derivative is calculated based on the spatial distribution of the phase distortion value. The corrected basic vector is subjected to directional enhancement filtering: the potential damage extension direction of the seam is retained to form a three-dimensional damage-sensitive vector field. The vector direction of each spatial point represents the sensitive direction of the damage response, and the vector modulus represents the response intensity of the position to the damage.

[0060] The working principle and beneficial effects of the above technical solution are as follows: First, the abnormal disturbance component output by the reverse decoupling of the multi-physical field is discretely sampled in the polarization direction according to a preset spatial grid in the entire longitudinal seam joint, and each sampling point captures the disturbance intensity and phase offset angle in a specific polarization plane; secondly, a three-dimensional spatial gradient operation is performed on the energy attenuation intensity of each sampling point in the abnormal disturbance component to generate an energy attenuation gradient vector field, the direction of which points to the direction of the maximum rate of change of energy attenuation, and the modulus characterizes the rate of change; the phase distortion value of each sampling point in the abnormal disturbance component is associated with the preset polarization direction to construct a polarization-phase correlation matrix, and the matrix elements quantify the contribution weight of the specific polarization direction to the phase distortion. Heavy; forming a dual-channel spatial field of coupled gradient and polarization properties, including an energy attenuation gradient vector field and a polarization-phase correlation matrix; finally, the direction of the energy attenuation gradient vector field in the dual-channel spatial field is used as the basic vector direction, the polarization-phase correlation matrix in the dual-channel spatial field is used to adjust the modulus of the basic vector according to the weight, and the second-order derivative of the phase distortion space is introduced to correct the basic vector direction. The second-order derivative is calculated by the spatial distribution of the phase distortion value; the corrected basic vector is subjected to directional enhancement filtering: the potential damage extension direction of the joint is retained to form a three-dimensional damage-sensitive vector field. The vector direction of each spatial point represents the sensitive direction of the damage response, and the vector modulus represents the response intensity of the position to the damage. The technical process of constructing a three-dimensional damage-sensitive vector field inside the longitudinal seam joint by the above scheme realizes the dynamic quantitative characterization of the structural damage evolution characteristics through the synergistic effect of multi-physical field inverse decoupling, spatial discrete sampling and vector field coupling operations. By simultaneously capturing the energy attenuation gradient and phase distortion polarization characteristics of anomalous perturbations through discrete sampling in the polarization direction, this method extends traditional single-field detection to a dual-channel coupled energy-phase analysis. The energy attenuation gradient vector field reflects the spatial distribution of material stiffness degradation, while the polarization-phase correlation matrix reveals the wave mode coupling effects caused by microscopic defects. Together, these two methods form a holographic representation of damage sensitivity through vector field operations. A vector direction correction mechanism based on the second-order spatial derivative of phase distortion overcomes the traditional gradient field's lack of sensitivity to damage propagation direction. A directional enhancement filtering algorithm, combined with joint geometry constraints, enables the resulting vector field to not only reflect the current damage state but also predict potential crack propagation paths through its vector topology. In the resulting three-dimensional vector field, the vector parameters (direction angle and modulus) at each spatial point strictly correspond to the physical response mechanism of material damage: the vector direction represents the direction of maximum damage sensitivity, and the modulus quantifies the degree of damage accumulation in a local region. This diagnostic method, based on the combined analysis of wave field polarization characteristics and energy dissipation, significantly improves the detection rate and localization accuracy of early-stage microdamage.

[0061] In summary, this embodiment transforms implicit material damage information into an explicit three-dimensional computable field by constructing a vector space mapping covering the full-band disturbance characteristics, thus providing a decision-making basis with clear physical meaning for structural health monitoring.

[0062] Example 4: Based on Example 3, the process of generating an energy attenuation gradient vector field provided in this embodiment of the present invention includes the following steps:

[0063] S2021: Input the discrete sampling result of the polarization direction, extract the energy attenuation intensity scalar value of each spatial grid point, and form a three-dimensional energy attenuation intensity discrete field covering the entire joint area. Each node of the three-dimensional energy attenuation intensity discrete field stores the energy loss rate in the specific polarization plane at that position;

[0064] S2022: With each discrete point in the three-dimensional energy attenuation intensity discrete field as the center, its differentiated neighborhood range is dynamically determined according to the joint geometric topology constraints; a large-scale neighborhood is used along the length of the joint to capture the long-range attenuation trend; a small-scale neighborhood is used perpendicular to the joint interface to capture the interface mutation characteristics; within the adaptive neighborhood, the energy attenuation intensity difference between the center point and each neighboring point is obtained to generate a spatial intensity difference set; a spatial intensity gradient feature set under geometric constraints is formed, which includes the adaptive neighborhood range of each discrete point and its corresponding intensity difference set;

[0065] S2023: Perform main direction screening on the intensity difference set of each discrete point in the spatial intensity gradient feature set, retain the neighborhood points whose absolute value of the difference is greater than the preset threshold, and eliminate noise interference; based on the screened neighborhood points, obtain the weighted vector sum of the intensity differences, and the weight is determined by the inverse of the spatial distance between the neighborhood point and the center point and the consistency coefficient of the seam direction. The vector sum direction is the direction of the maximum rate of change of energy attenuation, and the vector sum modulus is normalized to a quantized value of the change rate; form an energy attenuation gradient vector field, each discrete point corresponds to a vector, its direction represents the fastest enhancement path of energy attenuation, and its modulus represents the severity of attenuation.

[0066] The working principle and beneficial effects of the above technical solution are as follows: This embodiment first inputs the discrete sampling results of the polarization direction, extracts the energy attenuation intensity scalar value of each spatial grid point, and forms a three-dimensional energy attenuation intensity discrete field covering the entire joint. Each node of the three-dimensional energy attenuation intensity discrete field stores the energy loss rate in the specific polarization plane of the position; secondly, with each discrete point in the three-dimensional energy attenuation intensity discrete field as the center, its differentiated neighborhood range is dynamically determined according to the joint geometric topology constraint; a large-scale neighborhood is used along the length direction of the joint to capture the long-range attenuation trend; a small-scale neighborhood is used perpendicular to the joint interface to capture the interface mutation characteristics; in the adaptive neighborhood, the energy attenuation intensity difference between the center point and each neighborhood point is obtained to generate a spatial intensity difference set Combine; form a spatial intensity gradient feature set under geometric constraints, including the adaptive neighborhood range of each discrete point and its corresponding intensity difference set; finally, perform main direction screening on the intensity difference set of each discrete point in the spatial intensity gradient feature set, retain the neighborhood points whose absolute difference value is greater than the preset threshold, and eliminate noise interference; based on the screened neighborhood points, obtain the weighted vector sum of the intensity difference, the weight is determined by the inverse of the spatial distance between the neighborhood point and the center point and the consistency coefficient of the joint direction, the vector sum direction is the direction of the maximum rate of change of energy attenuation, and the vector sum modulus is normalized to the quantitative value of the rate of change; form an energy attenuation gradient vector field, each discrete point corresponds to a vector, its direction represents the fastest enhancement path of energy attenuation, and its modulus represents the severity of attenuation. The above scheme realizes the global quantitative representation of the energy attenuation characteristics of the weld joint and the visual reconstruction of the spatial evolution law through the gradient field construction method under multi-scale geometric constraints. Geometrically adaptive gradient field modeling: Through a dynamic neighborhood range determination mechanism, a macroscale neighborhood is used along the joint length to capture long-range energy transfer patterns, while a microscale neighborhood is used along the interface normal to analyze interface mutation behavior. This enables cross-scale coupled characterization of millimeter-scale structural features and micron-scale interface effects. Noise-robust gradient extraction: An intensity difference weighted algorithm based on principal direction screening effectively suppresses the interference of measurement noise on the gradient field direction by analyzing the spatial correlation of energy loss rates in the polarization plane, ensuring topological continuity in attenuation path identification. Physics-driven vector field construction: A two-parameter weighted model combining the inverse distance weight and the direction consistency coefficient ensures that the generated gradient vectors simultaneously satisfy: ① the spatial distance attenuation principle ② the joint geometry constraints, accurately reflecting the spatial distribution of energy attenuation rate and direction. Visual mapping of three-dimensional attenuation behavior: The process of converting discrete scalar fields into continuous vector fields provides a complete mathematical description of the spatial gradient of energy loss rate. The vector direction corresponds to the direction of maximum attenuation rate, and the modulus length represents the local energy dissipation intensity, providing a full-field quantitative basis for joint reliability assessment.

[0067] In summary, this embodiment forms an energy attenuation dynamics characterization system with clear physical meaning, which can support the three-dimensional attenuation behavior analysis needs in engineering application scenarios such as welding defect identification and life prediction.

[0068] Example 5: Figure 5 As shown, based on Example 1, the process of determining the dominant mode of latent damage and the critical state level of its development provided by the embodiment of the present invention includes the following steps:

[0069] S301: Input the generated three-dimensional damage-sensitive vector field, obtain the vector field time series data of the same joint area under preset time-varying detection conditions of different mechanical loads and temperature and humidity conditions, and form a dynamic vector field sequence, where each time slice contains the full-space vector distribution;

[0070] S302: Perform local manifold structure analysis on the vector field of each time slice in the three-dimensional damage-sensitive vector field, obtain the positions of three types of features: energy flow convergence points, divergence points, and vector curl anomaly areas, and integrate the corresponding vector modulus length mean and directional attributes into a spatiotemporal topological feature set; form a state matrix of time-varying topological features, where rows record different time points and columns record the spatial coordinates, intensity, direction angle, and area size of the three types of features;

[0071] S303: Perform intensity evolution trajectory convergence analysis on the spatial coordinate sequence of each type of feature in the state matrix, and calculate the standard deviation of the angle between the displacement vectors at adjacent time points; perform time domain gradient integration on the intensity sequence of each type of feature in the state matrix to obtain the intensity change per unit time; and form a multi-parameter evolution state diagnosis of hidden damage, including the dominant damage mode and critical level.

[0072] The working principle and beneficial effects of the above technical solution are as follows: This embodiment first inputs the generated three-dimensional damage-sensitive vector field and obtains time-series data of the vector field in the same joint area under preset time-varying detection conditions of different mechanical loads and temperature and humidity conditions, forming a dynamic vector field sequence, in which each time slice contains the full spatial vector distribution. Secondly, the vector field of each time slice in the three-dimensional damage-sensitive vector field is analyzed for local manifold structure, and the positions of three types of features (energy flow convergence points, divergence points, and vector curl anomaly areas) and the corresponding vector modulus mean and directional attributes are integrated into a spatiotemporal topological feature set. A state matrix of time-varying topological features is formed, with the state matrix rows recording different time points and the state matrix columns recording the spatial coordinates, intensity, directional angle, and area size of the three types of features. Finally, the spatial coordinate sequence of each type of feature in the state matrix is ​​analyzed for intensity evolution trajectory convergence, and the standard deviation of the angle between the displacement vectors at adjacent time points is calculated. The intensity sequence of each type of feature in the state matrix is ​​integrated into the time domain gradient to obtain the intensity change per unit time. This forms a multi-parameter evolution state diagnosis of hidden damage, including the dominant damage mode and critical level. The above scheme realizes the multi-dimensional dynamic diagnosis and critical state quantitative assessment of the hidden damage mode of the welded structure through the time-space coupled damage evolution analysis framework. Spatiotemporal correlation modeling of damage sensitive features: Based on the time-varying topological feature extraction of the dynamic vector field sequence, the spatial distribution of three types of damage sensitive features, namely energy flow convergence points, divergence points and vector curl anomaly areas, are correlated and mapped with the temporal evolution, and a full process characterization system for damage initiation, extension and critical failure is constructed. Automatic identification of damage dominant modes: Through the analysis of local manifold structure and the construction of time-varying state matrix, the characteristic spatial distribution laws corresponding to different damage modes (such as microcrack initiation, interface debonding, fatigue accumulation) are quantified, and unsupervised classification of damage types is realized. Dynamic prediction of critical states: Combining the convergence analysis of strength evolution trajectories with time-domain gradient integration, the nonlinear coupling relationship between the spatial migration rate of damage characteristics and the strength evolution trend is revealed. Through the joint criterion of the standard deviation of the displacement vector angle and the strength change, the state classification of damage from stable expansion (Level I) to accelerated expansion (Level II) to critical instability (Level III) is achieved; Multi-physics field coupled damage assessment: Through the time series comparison of vector fields under preset mechanical-thermal-humidity multi-field loading conditions, the evolutionary contribution of environmental factors and actual damage is separated, thereby improving the environmental robustness of latent damage diagnosis.

[0073] In summary, this embodiment establishes a full-chain monitoring capability from microscopic damage initiation to macroscopic failure, providing a dynamic evolution basis for active early warning of structural health status and remaining life prediction.

[0074] Example 6: Based on Example 5, the process of forming a state matrix of time-varying topological characteristics provided by the embodiment of the present invention includes the following steps:

[0075] S3021: Input the generated dynamic vector field sequence. In each time slice, the entire joint domain is divided into overlapping local manifold units. The unit size is constrained by the joint geometry to ensure coverage of the minimum damage scale.

[0076] S3022: Perform energy flow convergence point detection, divergence point detection, and vector curl anomaly area detection on each local manifold unit; bind the spatial position, intensity index, and dominant direction of the three types of features into a feature tuple; and form a time-space-strong coupling original feature set;

[0077] Energy flow convergence point detection: when all vectors in the local manifold unit satisfy the direction pointing to a certain internal focus and the modulus decreases with the distance from the focus, the coordinates of the focus and the mean modulus of the vectors in the unit are recorded; divergence point detection: when the center point of the local manifold unit satisfies the vector deviating from the center and the axial modulus along the seam increases with the distance, the coordinates of the center point and the axial modulus gradient value are recorded; vector curl anomaly area detection: calculate the maximum angle mutation value of the adjacent vector directions in the local manifold unit, and record the spatial range of the unit and the cumulative intensity of the angle mutation when it exceeds the threshold; the intensity index includes convergence point: mean modulus; divergence point: modulus gradient value; curl area: cumulative intensity of angle mutation; dominant direction includes the maximum mutation direction of the divergence point axial / curl area; each feature tuple of the original feature set of time-space-strong coupling contains timestamp, spatial coordinates, intensity index, and direction attribute;

[0078] S3023: All features in the original time-space-strong coupling feature set are classified and aggregated according to timestamp and feature type; multiple instances of each feature type at the same time are spatially clustered and merged; for the merged features, the spatial centroid coordinates, the mean of the merged strength index, and the composite value of the dominant direction are calculated; the merged features at each time point are filled with matrix rows according to a fixed dimension to form a state matrix of the time-varying topological features;

[0079] Among them, the convergence points / divergence points of spatial clustering merging: merge neighboring points whose Euclidean distance is less than the thickness of the seam; curl anomaly area: merge areas with overlapping boundaries; the dominant direction composite value is the vector average; the fixed dimensions include spatial coordinates x3, intensity x1, direction angle x3, and area size x1; the rows of the state matrix = time points, the columns = characteristic parameters, and the matrix elements are the numerical expressions of three-dimensional spatial coordinates / intensity / direction / size.

[0080] The working principle and beneficial effects of the above technical scheme are as follows: this embodiment first inputs the generated dynamic vector field sequence, and within each time slice, divides the entire joint into overlapping local manifold units, and the unit size is constrained by the joint geometric characteristics to ensure that the minimum damage scale is covered; secondly, energy flow convergence point detection, divergence point detection and vector curl anomaly area detection are performed on each local manifold unit; the spatial position, strength index and dominant direction of the three types of features are bound into feature tuples; a time-space-strong coupling original feature set is formed; finally, all features in the time-space-strong coupling original feature set are classified and aggregated according to timestamps and feature types; multiple instances of each type of feature at the same time are spatially clustered and merged; for the merged features, their spatial center of mass coordinates, the mean of the merged strength index and the composite value of the dominant direction are calculated; the merged features of each time point are filled with matrix rows according to a fixed dimension; a state matrix of time-varying topological features is formed; the above scheme constructs a structured characterization system for the evolution of hidden damage of welded joints through multi-scale feature fusion and time-space aggregation of dynamic vector fields. By dividing the local manifold units based on the constraints of the joint geometry, the scale adaptation of the macroscopic structural features and the microscopic damage signals is achieved. Based on the coordinated detection of energy flow convergence points, divergence points and vector curl anomaly areas, the three typical behavior modes of energy dissipation, stress concentration and material heterogeneity in the damage evolution process are fully captured. Using the time-space-strong coupling original feature set construction method, the spatial distribution, strength evolution and directional evolution of the three types of damage characteristics are uniformly parameterized to form a four-dimensional feature tuple containing timestamp, spatial coordinates, strength index and directional attributes, providing a multi-dimensional quantitative benchmark for dynamic damage monitoring. Through the spatial clustering merging algorithm based on the joint thickness, the intelligent fusion of multi-source damage features is achieved: the Euclidean distance constraint is implemented for the merging of discretely distributed convergence / divergence points, and the boundary overlap detection is implemented for the merging of continuously distributed curl anomaly areas. This process not only retains the true physical size of the damaged area but also eliminates detection redundancy. The final state matrix is ​​expressed through fixed-dimensional parameters, converting time-varying topological features into a computable data structure. Each element in the matrix corresponds to the numerical expression of the spatial coordinates, intensity, direction, and area size of a certain type of damage feature at a specific time point, providing standardized input for subsequent damage pattern recognition and critical state assessment.

[0081] In summary, this embodiment establishes a complete conversion chain from the original vector field to the structured state matrix, realizes the accurate mapping of the hidden damage characteristics from the physical space to the data space, and provides core data support for the quantitative diagnosis of the health status of the weld structure.

[0082] Example 7: Based on Example 5, the process of multi-parameter evolution state diagnosis of latent damage provided by the embodiment of the present invention includes the following steps:

[0083] S3031: Input the generated state matrix, whose row vectors contain the spatial coordinate sequence, intensity sequence, and direction angle sequence of three types of features: energy flow convergence point, divergence point, and vector curl anomaly area in time series;

[0084] S3032: For each type of feature's spatial coordinate sequence, calculate the coordinate differences between adjacent time points to form a displacement vector sequence. Perform standard deviation calculations on the direction angles of all adjacent displacement vectors in the spatial coordinate sequence to generate a path fluctuation coefficient. For each type of feature's intensity sequence, calculate the intensity differences between adjacent time points. Perform cumulative absolute value integration of the difference sequence along the time axis to generate an intensity transition cumulant. This forms a dual-channel parameter set for coupled spatial and intensity evolution, with each type of feature corresponding to one path fluctuation coefficient and one intensity transition cumulant.

[0085] S3033: If the path fluctuation coefficient of the divergence point is greater than the first preset threshold and its strength transition cumulative value is the largest, it is determined that the process is dominated by tensile debonding. If the strength transition cumulative value of the vector curl anomaly region and the path fluctuation coefficient of the energy flow convergence point increase synchronously, it is determined that the process is dominated by material degradation. Based on the relationship between the strength transition cumulative value and the preset threshold, the critical level is determined.

[0086] Critical state mapping:

[0087] When tensile debonding is dominant, the cumulative amount of strength transition is less than the second preset threshold, which is a stable latent period; when the cumulative amount of strength transition is between the second preset threshold and the third preset threshold, including the second preset threshold and the third preset threshold, it is a subcritical expansion period; when the cumulative amount of strength transition is greater than the third preset threshold, it is close to the instability period;

[0088] When material degradation is dominant: the path fluctuation coefficient of the energy flow convergence point is less than the fourth preset threshold value, which is a stable latent period; the ratio of the curl zone to the convergence point parameter is greater than the fifth preset threshold value, which is close to the instability period.

[0089] The working principle and beneficial effects of the above technical solution are as follows: This embodiment first inputs the generated state matrix, whose row vectors contain the spatial coordinate sequence, intensity sequence and direction angle sequence of three types of features: energy flow convergence point, divergence point and vector curl anomaly area in time series; secondly, for the spatial coordinate sequence of each type of feature, the coordinate difference of adjacent time points is calculated to form a displacement vector sequence, and the standard deviation operation is performed on the directional angles of all adjacent displacement vectors in the spatial coordinate sequence to generate a path fluctuation coefficient; for the intensity sequence of each type of feature, the intensity difference of adjacent time points is calculated, and the absolute value integral of the difference sequence is accumulated along the time axis to generate an intensity transition accumulation; a dual-channel parameter set of coupled space and intensity evolution is formed, and each type of feature corresponds to a path fluctuation coefficient and an intensity transition accumulation; finally, if the path fluctuation coefficient of the divergence point is greater than the first preset threshold and its intensity transition accumulation is the largest, it is determined to be dominated by tensile debonding; if the intensity transition accumulation of the vector curl anomaly area grows synchronously with the path fluctuation coefficient of the energy flow convergence point, it is determined to be dominated by material degradation; according to the intensity The critical level is determined by the relationship between the cumulative amount of strength transition and the preset threshold. Critical state mapping: When tensile debonding dominates, the cumulative amount of strength transition is less than the second preset threshold, indicating a stable latent period; when the cumulative amount of strength transition is between the second and third preset thresholds, inclusive, indicating a subcritical expansion period; when the cumulative amount of strength transition is greater than the third preset threshold, indicating an approaching instability period; when material degradation dominates: the energy flow convergence point path fluctuation coefficient is less than the fourth preset threshold, indicating a stable latent period; when the ratio of the curl zone to the convergence point parameter is greater than the fifth preset threshold, indicating an approaching instability period. The above scheme achieves cross-scale state mapping of hidden damage in shield tunnel longitudinal joints from microscopic evolution to macroscopic failure through multi-parameter dynamic coupling analysis. Damage mode decoupling and dominant mechanism identification: Through the dual-channel parameter coupling of path fluctuation coefficients and strength transition accumulation, the complex multi-physics perturbations at the joint interface are decomposed into two typical damage modes: tensile debonding (dominated by interface separation) and material degradation (dominated by bulk performance degradation). The directional dispersion of the displacement vector sequence reflects the directional instability of damage propagation, and the time-domain accumulation characteristics of intensity differences reveal the energy dissipation rate. These two factors work together to overcome the pattern confusion inherent in traditional single-parameter diagnosis. Dynamic calibration of criticality during the evolutionary stage: a multi-level criterion system based on parameter thresholds constructs a phase diagram of damage evolution: a stable latent period (damage initiation but no through-path formation), a subcritical propagation period (damage directionally develops but does not exceed structural tolerances), and a near-instability period (damage accumulation triggers nonlinear failure). This hierarchical mechanism accurately captures the transition from quantitative to qualitative change through the nonlinear growth characteristics of the strength transition accumulation and the spatial correlation of topological features (such as the parameter ratio of the curl zone to the convergence point).The spatiotemporal coupling mechanism of failure warning and the dynamic manifold analysis of the spatiotemporal coordinate sequence establish a ternary topological association between the damage core (energy flow convergence point), the expansion front (divergence point) and the performance degradation zone (vector curl anomaly area); by tracking the migration trajectory of characteristic points and the synchronization of intensity evolution, the transition threshold of damage from local concentration to global chain reaction is predicted in advance, providing time-varying failure boundary conditions for maintenance decisions.

[0090] In summary, this embodiment transforms the implicit material degradation process into a quantifiable dynamic parameter system through the inversion of the constitutive relationship between guided wave multimodal response and damage evolution, achieving a leap from phenomenological description to mechanistic prediction of structural health. The dominant mode relies on the coupled relationship between three characteristic parameters: tensile debonding is independently determined by a sudden increase in the divergence point parameter; material degradation is determined by the synchronization of the curl zone and convergence point parameters; the absolute value and relative relationship of the critical level mapping parameters: the first, second, third, and fourth preset thresholds are predetermined by the constitutive properties of the joint material; the parameter ratio (curl zone / convergence point) quantifies the correlation strength between material degradation and the damage core; and a fully probabilistic model: the diagnostic logic is constructed based on the deterministic relationship between physical parameters.

[0091] Example 8: Based on Example 7, the process of generating the path fluctuation coefficient and generating the intensity transition accumulation provided by the embodiment of the present invention includes the following steps:

[0092] S30321: Extract the spatial coordinate time series chain of the three-dimensional coordinates of each type of characteristic continuous time point and the intensity value time series chain of the intensity index value of each type of characteristic continuous time point from the input state matrix row vector, thereby forming the original sequence of characteristic spatiotemporal evolution;

[0093] S30322: For each feature's spatial coordinate time series chain in the original sequence of spatiotemporal evolution, the vector difference between the coordinates of adjacent time points is calculated to generate a displacement vector sequence, where each vector represents the spatial migration direction and distance of the feature within a unit of time. For each feature's intensity value time series chain in the original sequence of spatiotemporal evolution, the algebraic difference between the intensity values ​​at adjacent time points is calculated to generate an intensity difference sequence, where positive values ​​indicate enhancement and negative values ​​indicate weakening. This forms a dual-channel evolution field that couples spatial motion and intensity change. The displacement vector sequence describes the feature position evolution path, and the intensity difference sequence describes the feature intensity transition behavior.

[0094] Among them, based on the spatial coordinate time series chain in the original sequence of the characteristic spatiotemporal evolution, the three-dimensional coordinate components of adjacent time points are correspondingly subtracted, and the coordinates at the later time point are subtracted from the coordinates at the previous time point to generate a displacement vector sequence describing the direction and distance of the characteristic spatial migration; synchronously based on the intensity value time series chain, the intensity values ​​of adjacent time points are directly algebraically subtracted, and the intensity at the later time point is subtracted from the intensity at the previous time point to generate an intensity difference sequence reflecting the increase / decrease trend of the characteristic intensity; the time continuity feature, the displacement vector sequence provides the direction analysis basis for the path fluctuation coefficient, and the intensity difference sequence provides the mutation quantification basis for the accumulated amount of intensity transition. The two together constitute the core elements of the dual-channel evolution field;

[0095] S30323: Extract the directional unit vectors of all adjacent displacement vectors from the displacement vector sequence of the dual-channel evolution field, calculate the dot product arccosine value of each pair of adjacent directional vectors, find the standard deviation of the dot product arccosine value, and output the path fluctuation coefficient; take the absolute value of the intensity difference sequence of the dual-channel evolution field, perform linear weighted accumulation on the absolute value sequence along the time axis, and output the intensity transition accumulation; form a dual-channel evolution parameter set, where each feature corresponds to two parameters: the path fluctuation coefficient and the intensity transition accumulation.

[0096] The working principle and beneficial effects of the above technical solution are as follows: This embodiment first inputs the row vector of the state matrix, extracts the spatial coordinate time series chain of the three-dimensional coordinates of each type of feature continuous time point and the intensity value time series chain of the intensity index value of the continuous time point; forms the original sequence of feature spatiotemporal evolution; secondly, for the spatial coordinate time series chain of each feature in the original sequence of feature spatiotemporal evolution, calculates the vector difference of the coordinates of adjacent time points, generates a displacement vector sequence, each vector represents the spatial migration direction and distance of the feature in unit time; for the intensity value time series chain of each feature in the original sequence of feature spatiotemporal evolution, calculates the algebraic difference of the intensity values ​​of adjacent time points, generates an intensity difference sequence, and a positive value indicates enhancement , negative values ​​indicate weakening; a dual-channel evolution field is formed that couples spatial motion and intensity changes, with the displacement vector sequence describing the feature position evolution path and the intensity difference sequence describing the feature intensity transition behavior; finally, the directional unit vectors of all adjacent displacement vectors are extracted from the displacement vector sequence of the dual-channel evolution field, the dot product arccosine of each pair of adjacent directional vectors is calculated, the standard deviation of the dot product arccosine values ​​is calculated, and the path fluctuation coefficient is output; the absolute value of the intensity difference sequence of the dual-channel evolution field is taken, and the absolute value sequence is linearly weighted and accumulated along the time axis to output the intensity transition accumulation; a dual-channel evolution parameter set is formed, with each feature corresponding to two parameters: the path fluctuation coefficient and the intensity transition accumulation. The above scheme achieves quantitative characterization of dynamic feature evolution behavior through spatiotemporal coupling analysis; a dual-channel evolution field is constructed to synchronously capture the spatiotemporal coupling relationship between the spatial displacement vector and the intensity difference sequence, and then high-order feature parameters are extracted through vector operations and sequence analysis. The path fluctuation coefficient quantifies the frequency and amplitude of directional mutations in characteristic motion trajectories by performing the arccosine dot product of the directional unit vectors of the displacement vector sequence. Its standard deviation eliminates dimension effects and accurately reflects the tortuosity and uncertainty of the path. This parameter overcomes the traditional Euclidean distance metric's inability to capture directional changes. The intensity transition cumulant employs an absolute value linear weighted accumulation strategy, preserving the original polarity information of the intensity changes (via pre-processing of the difference sequence) while highlighting the total amount of transitions through absolute value conversion. Weighted accumulation emphasizes the contribution of recent changes. This method effectively distinguishes between distinct evolutionary patterns, such as sustained enhancement and intermittent fluctuations. The synergistic output of the two-channel parameters constructs a complete spatiotemporal evolution feature space: the path fluctuation coefficient describes the geometric characteristics of the motion trajectory, while the intensity transition cumulant represents the energy transfer efficiency. Both maintain physical consistency through the generation mechanism of the original sequence aligned in time and space. This coupled quantification approach improves the granularity of analysis for complex evolutionary behaviors compared to single-dimensional time series analysis.

[0097] In summary, this embodiment achieves the computable transformation of nonlinear evolution processes through the third-order processing of spatiotemporal chain decomposition-vector differential operation-multi-channel statistical modeling, providing a physically interpretable feature engineering framework for scenarios such as stability analysis of dynamic systems and abnormal pattern detection.

[0098] Example 9: Based on Example 8, the process of outputting the path fluctuation coefficient provided in this embodiment of the present invention includes the following steps:

[0099] S303231: Input the generated displacement vector sequence and perform directional normalization on each displacement vector in the sequence to form a set of directional unit vectors of the displacement vector sequence. Each vector represents the pure direction of migration of the feature in unit time.

[0100] S303232: For the time series chain of directional unit vectors in the directional unit vector set, take each pair of adjacent vectors in chronological order, calculate the dot product of each pair of adjacent vectors, and perform an inverse cosine function on the dot product to generate a sequence of directional deflection angles. Each angle value quantifies the instantaneous deflection amplitude of the migration direction in adjacent time periods. This forms a continuous record of directional deflection disturbances. The angle sequence directly reflects the severity of the path direction change.

[0101] S303233: Perform global statistical dispersion analysis on the directional deflection angle sequence in the continuous directional deflection disturbance record, calculate the arithmetic mean of all angles in the directional deflection angle sequence, calculate the sum of the squares of the absolute deviations of each angle from the mean, and take the root mean square operation on the sum of squares; form the path fluctuation coefficient. The larger the value of the path fluctuation coefficient, the greater the disorder of the migration direction and the lower the path stability.

[0102] The working principle and beneficial effects of the above technical solution are as follows: This embodiment first inputs the generated displacement vector sequence, performs directional normalization on each displacement vector in the sequence, and forms a set of directional unit vectors of the displacement vector sequence, each vector representing the pure direction of migration of the feature within a unit time. Secondly, for the directional unit vector time series chain in the directional unit vector set, each pair of adjacent vectors is taken in chronological order, the dot product value of each pair of adjacent vectors is calculated, and the inverse cosine function is applied to the dot product value to generate a sequence of directional deflection angles, each angle value quantifies the instantaneous deflection amplitude of the migration direction in adjacent time periods. A continuous directional deflection disturbance record is formed, and the angle sequence directly reflects the severity of the path direction change. Finally, the directional deflection angle sequence in the continuous directional deflection disturbance record is subjected to global statistical dispersion analysis, and the arithmetic mean of all angles in the directional deflection angle sequence is calculated. The sum of the absolute squares of the deviations of each angle from the mean is calculated, and the root mean square operation is performed on the sum of squares. The path fluctuation coefficient is formed. The larger the path fluctuation coefficient value, the greater the disorder of the migration direction and the lower the path stability. The above scheme achieves quantitative characterization of the directional stability of the motion trajectory by integrating vector differential geometry and statistical dynamics methods. A multi-order analysis of directional perturbation decomposes a continuous motion trajectory into a sequence of instantaneous directional deflections by normalizing the displacement vector and performing the arccosine operation on the dot product of adjacent vectors. This overcomes the traditional curvature calculation's reliance on the continuous path assumption and directly captures the directional abrupt changes at discrete time scales. Statistical modeling of path disorder: Based on the root mean square (RMS) dispersion of the directional deflection angle sequence, a rotationally invariant fluctuation coefficient is constructed. Its physical meaning is the energy density of the directional perturbation. This coefficient amplifies anomalous deflection events in non-stationary motion through a second-order statistic (RMS of the sum of squared deviations) and offers strong discrimination against non-deterministic paths such as random walks and Brownian motion. Completeness of stability assessment: The final output path fluctuation coefficient combines the instantaneous amplitude of directional changes (ARC operation) with long-term statistical characteristics (RMS), reflecting both the local impact of a single sharp turn and the degree of anisotropy of the overall motion pattern. This joint time-frequency analysis makes it superior to traditional methods based solely on curvature integrals or Fourier transforms.

[0103] In summary, this embodiment transforms geometrically intuitive directional changes into stability indicators with strict probabilistic interpretation through the cascade of normalization, dot product, arccosine, and root mean square discreteness, providing differentiable high-order features for tasks such as motion trajectory classification and phase change detection in dynamic systems.

[0104] Example 10: Based on Example 9, the method of calculating the dot product value of each pair of adjacent vectors provided in this embodiment of the present invention includes the following steps:

[0105] S3032321: Input the time series chain of directional unit vectors, extract all consecutive moment vector combinations in the order of detection time, and generate time series adjacent directional vector pairs;

[0106] S3032322: Execute for each pair of adjacent directional vectors in the time series, multiplying the three-dimensional components of the two vectors to generate a directional consistency scalar value that represents the spatial overlap of the two vector directions. This forms a directional consistency scalar sequence, where each scalar value uniquely quantifies the continuity of the migration direction in adjacent time periods.

[0107] S3032323: Use the scalar sequence in the direction consistency scalar sequence as the input parameter for the direction deflection angle calculation to determine the output range of the arc cosine operation; and form a dot product value sequence.

[0108] The working principle and beneficial effects of the above technical solution are as follows: This embodiment first inputs a time-series chain of directional unit vectors, extracts all continuous moment vector combinations in the order of detection time, and generates time-series adjacent directional vector pairs; secondly, it executes each adjacent vector pair in the time-series adjacent directional vector pairs, multiplies the three-dimensional components of the two vectors correspondingly, generates a directional consistency scalar value, and characterizes the spatial coincidence of the directions of the two vectors; forms a directional consistency scalar sequence, each scalar value uniquely quantifies the continuity of the migration direction of adjacent time periods; finally, the scalar sequence in the directional consistency scalar sequence is used as the direction The input parameters for the deflection angle calculation determine the output range of the inverse cosine operation, forming a sequence of dot product values. This dot product is obtained by multiplying the three-dimensional components of adjacent unit direction vectors. This scalar sequence of directional consistency is strictly confined to the interval [-1, 1], directly corresponding to the cosine of the spatial angle between the two vectors. When this sequence is input to the inverse cosine function, its range naturally constrains the output angle to the range [0, π] radians. This transforms the problem of geometric directional continuity into a differentiable scalar operation chain, ensuring the physical meaning of the directional deflection angle is clear and computationally stable. This scheme achieves precise quantification of motion directional continuity through the coordinated processing of vector algebraic operations and geometric transformations. Frame-by-frame analysis of directional consistency: By performing component-by-component dot products on pairs of adjacent vectors in time series, directional changes in three-dimensional space are converted into a scalar sequence. This preserves the spatial angle between the vectors while achieving a balance between computational efficiency and geometric meaning through scalarization. This overcomes the singularity problem associated with direct Euler angle calculations and is applicable to motion trajectory analysis in any dimension. Differentiable representation of dynamic continuity: The generated directional consistency scalar sequence is essentially a record of the time evolution of direction cosines. Its value range (-1 to 1) strictly corresponds to the physical constraints of the vector angle, with negative values ​​representing direction reversals, zero representing orthogonal mutations, and positive values ​​representing gradual turns. The normalized output provides a mathematically complete input space for subsequent inverse cosine operations. Structural guarantees of computational stability: By enforcing temporal proximity constraints and deterministic operations such as corresponding component multiplication, the common problem of cumulative error in trajectory analysis is avoided. The scalar sequence, as an intermediate variable, isolates the original data from noise while maintaining the reversibility of geometric transformations through linear operations, providing a numerically stable foundation for high-order feature extraction.

[0109] In summary, this embodiment transforms the directional continuity detection in kinematics into a linear algebra problem with strict mathematical definition by establishing a cascade data processing chain of vector pairs, scalar sequences, and arccosine inputs. Its output can directly support stability analysis based on differential geometry or pattern classification based on statistical mechanics.

[0110] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the present invention's equivalent technologies, the present invention is also intended to include these modifications and variations.

Claims

1. A method for diagnosing hidden damage of longitudinal seam joints in shield tunnels based on multimodal ultrasonic guided waves, characterized in that: The method includes the following steps: using the generated three-dimensional damage-sensitive vector field to perform dynamic manifold topological analysis, identifying the topological features of the three-dimensional damage-sensitive vector field, including the energy flow convergence points of potential damage cores, the divergence points of the damage extension front, and the vector curl anomaly areas of the material performance degradation zone, and tracking the migration paths and strength evolution trajectories of the topological features as the detection conditions change; and determining the dominant mode of latent damage and the critical state level of its development based on the migration path stability and strength evolution rate of the topological features. The process of generating a three-dimensional damage-sensitive vector field includes the following steps: using the acquired guided wave interferometer pattern and performing multi-physics inverse decoupling to separate the baseline interference characteristics caused by the joint geometric discontinuity and the abnormal disturbance component caused by the hidden damage in the guided wave interferometer pattern. The abnormal disturbance component manifests as abnormal energy attenuation and phase distortion in a specific polarization direction; forming the spatial distribution and intensity of the abnormal disturbance component based on the multi-physics inverse decoupling, constructing a three-dimensional damage-sensitive vector field inside the longitudinal seam joint, and quantifying the response intensity and directionality of different locations to the damage pattern; The process of generating the guided wave interferometry pattern includes the following steps: applying a composite ultrasonic energy field with orthogonal polarization, including longitudinal and shear wave energy flows with a specific frequency combination, to a specific area of ​​the longitudinal joint; and forming a spatiotemporally coupled guided wave interferometry pattern over the entire longitudinal joint, including the joint body and the concrete on both sides. The guided wave interferometry pattern includes the amplitude distribution, phase delay, and modal conversion characteristics under the interaction of the composite ultrasonic energy field.

2. The method for diagnosing hidden damage of longitudinal seam joints in shield tunnels based on multimodal ultrasonic guided waves according to claim 1, characterized in that: The process of determining the dominant mode of latent damage and the criticality level of its development includes the following steps: The generated three-dimensional damage-sensitive vector field is input and the vector field time series data of the same joint area is obtained under the time-varying detection conditions of different preset mechanical loads and temperature and humidity states to form a dynamic vector field sequence. Each time slice contains the full space vector distribution. The local manifold structure of the vector field of each time slice in the three-dimensional damage-sensitive vector field is analyzed. The positions of three types of features, namely energy flow convergence points, divergence points, and vector curl anomaly areas, and the corresponding vector modulus mean and directional attributes are integrated into a spatiotemporal topological feature set. The state matrix of the time-varying topological features is formed. The rows of the state matrix record different time points, and the columns record the spatial coordinates, intensity, direction angle, and regional size of the three types of features. The spatial coordinate sequence of each type of feature in the state matrix is ​​subjected to a convergence analysis of the intensity evolution trajectory, and the standard deviation of the angles between the displacement vectors at adjacent time points is calculated. The intensity sequence of each type of feature in the state matrix is ​​subjected to a time-domain gradient integration to obtain the intensity change per unit time. This forms a multi-parameter evolution state diagnosis of latent damage, including the dominant damage mode and critical level.

3. The method for diagnosing hidden damage of longitudinal seam joints in a shield tunnel based on multimodal ultrasonic guided waves according to claim 2, characterized in that: The process of forming a state matrix of time-varying topological features includes the following steps: The generated dynamic vector field sequence is input and in each time slice, the entire joint domain is divided into overlapping local manifold cells. The cell size is constrained by the joint geometry characteristics to ensure that the minimum damage scale is covered. Energy flow convergence point detection, divergence point detection, and vector curl anomaly area detection are performed on each local manifold unit; the spatial position, intensity index, and dominant direction of the three types of features are bound into a feature tuple to form a time-space-strong coupling original feature set; All features in the original time-space-strong coupling feature set are classified and aggregated by timestamp and feature type; Multiple instances of each type of feature at the same time are spatially clustered and merged; for the merged features, their spatial centroid coordinates, the mean of the merged strength index and the composite value of the dominant direction are calculated; the merged features at each time point are filled with matrix rows according to a fixed dimension; and a state matrix of time-varying topological features is formed.

4. The method for diagnosing hidden damage of longitudinal seam joints in a shield tunnel based on multimodal ultrasonic guided waves according to claim 2, characterized in that: The process of multi-parameter evolution state diagnosis of hidden damage includes the following steps: The generated state matrix is ​​input, and its row vectors contain the spatial coordinate sequence, intensity sequence, and direction angle sequence of three types of features: energy flow convergence point, divergence point, and vector curl anomaly area in time series; For each type of feature's spatial coordinate sequence, the coordinate differences between adjacent time points are calculated to form a displacement vector sequence. The standard deviation of the direction angles of all adjacent displacement vectors in the spatial coordinate sequence is calculated to generate the path fluctuation coefficient. For each type of feature's intensity sequence, the intensity difference between adjacent time points is calculated. The difference sequence is then accumulated along the time axis by performing an absolute value integration to generate the intensity transition accumulation. A dual-channel parameter set of coupled space and intensity evolution is formed, where each type of feature corresponds to a path fluctuation coefficient and an intensity transition accumulation; If the path fluctuation coefficient of the divergent point is greater than the first preset threshold and the cumulative amount of its strength transition is the largest, it is determined that the tensile debonding is dominant; If the cumulative amount of intensity transition in the vector curl anomaly area increases synchronously with the path fluctuation coefficient of the energy flow convergence point, it is determined that material degradation is dominant; the critical level is determined based on the relationship between the cumulative amount of intensity transition and the preset threshold.

5. The method for diagnosing hidden damage of longitudinal seam joints in a shield tunnel based on multimodal ultrasonic guided waves according to claim 4, characterized in that: Critical state mapping: When tensile debonding is dominant, the cumulative amount of strength transition is less than the second preset threshold, which is a stable latent period; when the cumulative amount of strength transition is between the second preset threshold and the third preset threshold, including the second preset threshold and the third preset threshold, it is a subcritical expansion period; when the cumulative amount of strength transition is greater than the third preset threshold, it is close to the instability period; When material degradation is dominant: the path fluctuation coefficient of the energy flow convergence point is less than the fourth preset threshold value, which is a stable latent period; the ratio of the curl zone to the convergence point parameter is greater than the fifth preset threshold value, which is close to the instability period.

6. The method for diagnosing hidden damage of longitudinal seam joints in a shield tunnel based on multimodal ultrasonic guided waves according to claim 4, characterized in that: The process of generating the path fluctuation coefficient and the intensity transition accumulation includes the following steps: The input state matrix row vector is used to extract the spatial coordinate time series chain of the three-dimensional coordinates of each type of characteristic continuous time point and the intensity value time series chain of the intensity index value of the continuous time point; thus forming the original sequence of the characteristic spatiotemporal evolution; For the spatial coordinate time series chain of each feature in the original sequence of the spatiotemporal evolution of the features, the vector difference of the coordinates of adjacent time points is calculated to generate a displacement vector sequence; for the intensity value time series chain of each feature in the original sequence of the spatiotemporal evolution of the features, the algebraic difference of the intensity values ​​of adjacent time points is calculated to generate an intensity difference sequence; a dual-channel evolution field of coupled spatial motion and intensity change is formed, in which the displacement vector sequence describes the evolution path of the feature position, and the intensity difference sequence describes the transition behavior of the feature intensity; The directional unit vectors of all adjacent displacement vectors are extracted from the displacement vector sequence of the dual-channel evolution field, the dot product arc cosine value of each pair of adjacent directional vectors is calculated, the standard deviation of the dot product arc cosine value is obtained, and the path fluctuation coefficient is output; the absolute value of the intensity difference sequence of the dual-channel evolution field is taken, the absolute value sequence is linearly weighted accumulated along the time axis, and the intensity transition accumulation is output; a dual-channel evolution parameter set is formed, and each feature corresponds to two parameters: the path fluctuation coefficient and the intensity transition accumulation.

7. The method for diagnosing hidden damage of longitudinal seam joints in a shield tunnel based on multimodal ultrasonic guided waves according to claim 6, characterized in that: Each vector in the displacement vector sequence represents the spatial migration direction and distance of the feature in unit time; the positive value of the intensity difference sequence indicates enhancement, and the negative value indicates weakening.

8. The method for diagnosing hidden damage of longitudinal seam joints in a shield tunnel based on multimodal ultrasonic guided waves according to claim 1, characterized in that: The process of constructing the three-dimensional damage sensitivity vector field inside the longitudinal seam joint includes the following steps: The abnormal disturbance component output by the reverse decoupling of the multi-physics field is discretely sampled in the polarization direction according to a preset spatial grid in the entire longitudinal seam joint. Each sampling point captures the disturbance intensity and phase offset angle in a specific polarization plane. Perform three-dimensional spatial gradient calculation on the energy attenuation intensity of each sampling point in the abnormal disturbance component to generate an energy attenuation gradient vector field; The phase distortion value of each sampling point in the abnormal disturbance component is associated with the preset polarization direction to construct a polarization-phase correlation matrix; thus forming a dual-channel spatial field with coupled gradient and polarization properties; The direction of the energy attenuation gradient vector field in the dual-channel spatial field is taken as the direction of the basis vector. The modulus of the basis vector is adjusted according to the weight of the polarization-phase correlation matrix in the dual-channel spatial field. The second-order derivative of the phase distortion space is introduced to correct the direction of the basis vector. The corrected basis vector is subjected to directional enhancement filtering to retain the potential damage extension direction of the seam, forming a three-dimensional damage-sensitive vector field.

Images