Multi-modal ultrasonic guided wave-based shield tunnel longitudinal seam joint hidden damage diagnosis method

Abstract

The invention provides a shield tunnel longitudinal seam joint recessive damage diagnosis method based on multi-mode ultrasonic guided waves, and the method comprises the steps: applying orthogonal polarization in a specific region of a longitudinal seam joint, and forming a space-time coupled guided wave interference pattern in the whole region of the longitudinal seam joint; the acquired guided wave interference map is utilized to separate baseline interference characteristics and abnormal disturbance components caused by hidden damage through multi-physics field reverse decoupling, and the abnormal disturbance components are represented as energy abnormal attenuation and phase distortion in a specific polarization direction; constructing a three-dimensional damage sensitive vector field in the longitudinal seam joint; and carrying out dynamic manifold topology analysis by utilizing the generated three-dimensional damage sensitive vector field, identifying topological characteristics in the three-dimensional damage sensitive vector field, and tracking a migration path and an intensity evolution trajectory of the topological characteristics along with the change of the detection condition. According to the method, full-dimension diagnosis and evolution prediction of the hidden damage of the longitudinal joint connector of the shield tunnel are realized.

Description

Technical Field

[0001] The present invention relates to the technical field of longitudinal joint damage judgment, and particularly relates to a method for diagnosing hidden damage of longitudinal joints in shield tunnels based on multimodal ultrasonic guided waves. Background Art

[0002] The expansion of modern transportation networks has led to a sharp increase in tunnel construction volume. As the core connecting component of precast concrete segments, the evolution of hidden damage in longitudinal joints directly threatens the service safety of structures. The current industry faces three dilemmas: the problem of concealment, where damage such as internal steel bar corrosion and concrete microcracks in longitudinal joints is wrapped by a waterproof sealing layer, and the failure rate of traditional visual inspection reaches 67%. The time-delay effect, with an average latency period of 8 - 12 months from damage initiation to visible surface cracks. The chain risk, where damage to a single longitudinal joint may cause stress redistribution in adjacent segments, leading to domino-style structural degradation.

[0003] Prior art one, application number: 202311008097.6 discloses a method and device for ultrasonic detection of the health status of longitudinal joints in operating shield tunnels. The method includes: starting a concrete ultrasonic detector and recording the acoustic time when ultrasonic waves propagate from the transmitting probe to the receiving probe; keeping the transmitting probe stationary, each time moving the receiving probe away from the longitudinal joint at equal distances, starting the concrete ultrasonic detector, and successively recording the acoustic time when ultrasonic waves propagate from the transmitting probe to the receiving probe at each distance; exchanging the positions of the transmitting probe and the receiving probe, fixing the transmitting probe on one side near the longitudinal joint edge of the original receiving probe and keeping it stationary, and placing the receiving probe on the original transmitting probe side, repeating the measurement of acoustic time at different distances. The distance between the receiving probe and the longitudinal joint edge when the acoustic time is the smallest is the length of the fracture crack, and the damage degree of the longitudinal joint is judged by the length of the fracture crack. The device includes: an ultrasonic detector, a transmitting probe, a receiving probe, and a longitudinal joint. Although it can provide effective data support for the health evaluation of current operating shield tunnels; however, it only relies on acoustic time measurement: using single-probe ultrasonic detection, only recording the ultrasonic propagation time, indirectly judging the damage degree through the length of the fracture crack, and unable to accurately identify hidden damage; single detection dimension: only performing one-dimensional evaluation based on the change of acoustic time, unable to reflect the spatial distribution, modal conversion characteristics, and damage evolution trend of damage; weak anti-interference ability: greatly affected by factors such as the anisotropy of concrete materials and environmental noise, and it is difficult to distinguish geometric discontinuities from real damage; unable to achieve three-dimensional damage characterization: only able to provide simple parameters such as crack length, and unable to construct a vector field or topological evolution model of damage.

[0004] Prior Art 2, Application No.: 201010289974.8 discloses a mechanical property test system for a multi-functional lining segment joint capable of three-way loading, which mainly includes a self-balanced frame subsystem, a loading subsystem, a specimen support subsystem, a specimen transportation subsystem, a loading control subsystem, and a data acquisition subsystem. The loading subsystem adopts a modular design and can complete shear tests between segments of a shield tunnel lining, moment transfer tests, longitudinal joint angular stiffness tests, longitudinal joint radial shear tests, etc. through different combinations. Although it can better simulate the three stress states of a shield tunnel lining segment joint under actual conditions, can complete various different loading methods to measure the mechanical parameters of the shield tunnel lining segment joint, as well as the mechanical properties of other similar structures; however, it is only applicable to laboratory loading tests: relying on mechanical loading to simulate the stress state, it cannot be directly used for on-site non-destructive testing; it cannot identify hidden damages: it can only measure macroscopic mechanical parameters and cannot detect early damages such as microcracks and material degradation; non-dynamic monitoring: it belongs to static or quasi-static tests and cannot track the damage propagation process in real time.

[0005] Currently, Prior Art 1 and Prior Art 2 have problems such as being unable to accurately identify hidden damages, lacking three-dimensional dynamic damage characterization capabilities, and poor anti-interference performance. Therefore, the present invention provides a method for diagnosing hidden damages in longitudinal joints of shield tunnels based on multi-modal ultrasonic guided waves. Summary of the Invention

[0006] To solve the above technical problems, the present invention provides a method for diagnosing hidden damages in longitudinal joints of shield tunnels based on multi-modal ultrasonic guided waves, which includes the following steps: using the generated three-dimensional damage-sensitive vector field, performing dynamic manifold topology analysis, identifying the topological features of the energy flow convergence points containing potential damage cores, the divergence points of the damage propagation frontiers, and the vector curl anomaly regions of the material property degradation zones in the three-dimensional damage-sensitive vector field, and tracking the migration paths and intensity evolution trajectories of the topological features with changes in detection conditions; according to the stability of the migration paths of the topological features and the intensity evolution rate, determining the dominant mode of the hidden damage and the critical state level at which its development is located.

[0007] Optionally, the process of determining the dominant mode of the hidden damage and the critical state level at which its development is located includes the following steps: Input the generated three-dimensional damage-sensitive vector field, obtain the time series data of the vector field in the same joint area under time-varying detection conditions with preset different mechanical loads and temperature and humidity states, form a dynamic vector field sequence, and each time slice contains the vector distribution of the entire space; Perform local manifold structure analysis on the vector field of each time slice in the three-dimensional damage-sensitive vector field, and integrate the positions of three types of features, namely energy flow convergence points, divergence points, and vector curl anomaly regions, and the corresponding mean vector modulus lengths and direction attributes into a spatio-temporal topological feature set; form a state matrix of time-varying topological features, where the rows of the state matrix record different time points, and the columns record the spatial coordinates, intensities, direction angles, and regional sizes of the three types of features; Perform convergence analysis on the intensity evolution trajectories of the spatial coordinate sequences of each type of feature in the state matrix, and calculate the standard deviation of the included angles of the displacement vectors at adjacent time points; perform time-domain gradient integration on the intensity sequences of each type of feature in the state matrix to obtain the intensity change amount per unit time; form a multi-parameter evolution state diagnosis of hidden damage, including the dominant damage mode and critical level.

[0008] Optionally, the process of forming the state matrix of time-varying topological features includes the following steps: Input the generated dynamic vector field sequence. Within each time slice, divide the entire joint area into overlapping local manifold units, and the unit size is constrained by the seam geometric features to ensure coverage of the minimum damage scale; Perform energy flow convergence point detection, divergence point detection, and vector curl anomaly region detection on each local manifold unit; bind the spatial positions, intensity indicators, and dominant directions of the three types of features into feature tuples; form a raw feature set of time-space-intensity coupling; Classify and aggregate all features in the raw feature set of time-space-intensity coupling according to time stamps and feature types; perform spatial clustering and merging on multiple instances of each type of feature at the same moment; for the merged features, calculate their spatial centroid coordinates, the mean value of the merged intensity indicators, and the synthesized value of the dominant direction; fill the matrix rows with the merged features at each time point according to a fixed dimension; form the state matrix of time-varying topological features.

[0009] Optionally, the process of forming a multi-parameter evolution state diagnosis of hidden damage includes the following steps: Input the generated state matrix, whose row vectors include the spatial coordinate sequences, intensity sequences, and direction angle sequences of three types of features, namely energy flow convergence points, divergence points, and vector curl anomaly regions, in time series; For the spatial coordinate sequence of each type of feature, calculate the coordinate differences at adjacent time points to form a displacement vector sequence, perform a standard deviation operation on the direction 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, calculate the intensity differences at adjacent time points, and perform cumulative absolute value integration on the difference sequence along the time axis to generate an intensity jump cumulative amount; form a two-channel parameter set that couples spatial and intensity evolution, with each type of feature corresponding to a path fluctuation coefficient and an intensity jump cumulative amount; If the path fluctuation coefficient of the divergence point is greater than the first preset threshold and its intensity transition cumulative amount is the largest, it is determined to be dominated by tensile debonding; if the intensity transition cumulative amount in the abnormal vector curl region and the path fluctuation coefficient of the energy flow convergence point increase synchronously, it is determined to be dominated by material degradation; according to the relationship between the intensity transition cumulative amount and the preset threshold, the critical level is judged.

[0010] Optionally, critical state mapping: When it is dominated by tensile debonding, if the intensity transition cumulative amount is less than the second preset threshold, it is the stable latency period; if the intensity transition cumulative amount is between the second preset threshold and the third preset threshold, including the second preset threshold and the third preset threshold, it is the subcritical expansion period; if the intensity transition cumulative amount is greater than the third preset threshold, it is the near instability period; When it is dominated by material degradation: if the path fluctuation coefficient of the energy flow convergence point is less than the fourth preset threshold, it is the stable latency period; if the ratio of the curl region to the convergence point parameter is greater than the fifth preset threshold, it is the near instability period.

[0011] Optionally, the process of generating the path fluctuation coefficient and the intensity transition cumulative amount includes the following steps: For the input state matrix row vector, extract the spatial coordinate time series chain of the three-dimensional coordinates of each type of feature at consecutive time points and the intensity value time series chain of the intensity index values at consecutive time points; form the original feature spatio-temporal evolution sequence; For the spatial coordinate time series chain of each feature in the original feature spatio-temporal evolution sequence, calculate the vector difference of the coordinates at adjacent time points to generate a displacement vector sequence; for the intensity value time series chain of each feature in the original feature spatio-temporal evolution sequence, calculate the algebraic difference of the intensity values at adjacent time points to generate an intensity difference sequence; form a two-channel evolution field coupling spatial motion and intensity change, where the displacement vector sequence describes the evolution path of the feature position and the intensity difference sequence describes the intensity transition behavior of the feature; Extract the direction unit vectors of all adjacent displacement vectors from the displacement vector sequence in the two-channel evolution field, calculate the arccosine value of the dot product of each pair of adjacent direction vectors, calculate the standard deviation of the arccosine values of the dot products, and output the path fluctuation coefficient; take the absolute value of the intensity difference sequence in the two-channel evolution field and perform linear weighted accumulation along the time axis to output the intensity transition cumulative amount; form a two-channel evolution parameter set, with each feature corresponding to two parameters: the path fluctuation coefficient and the intensity transition cumulative amount.

[0012] Optionally, each vector in the displacement vector sequence represents the spatial migration direction and distance of the feature within a unit time; a positive value in the intensity difference sequence indicates enhancement, and a negative value indicates weakening.

[0013] Optionally, the generation process of the three-dimensional damage-sensitive vector field includes the following steps: Using the obtained guided wave interference pattern, through multi-physical field reverse decoupling, separating the baseline interference characteristics generated by the joint geometric discontinuity in the guided wave interference pattern and the abnormal perturbation components caused by hidden damage. The abnormal perturbation components are manifested as abnormal energy attenuation and phase distortion in a specific polarization direction; forming the spatial distribution and intensity of the abnormal perturbation components reverse decoupled by multi-physical fields, constructing the three-dimensional damage-sensitive vector field inside the longitudinal joint, and quantifying the response intensity and directionality of different positions to damage modes.

[0014] Optionally, the process of constructing the three-dimensional damage-sensitive vector field inside the longitudinal joint includes the following steps: Performing polarization direction discrete sampling on the abnormal perturbation components output by multi-physical field reverse decoupling in the entire longitudinal joint area according to a preset spatial grid, and each sampling point captures the perturbation intensity and phase offset angle in a specific polarization plane; Performing three-dimensional spatial gradient operation on the energy attenuation intensity of each sampling point in the abnormal perturbation components to generate an energy attenuation gradient vector field; correlating the phase distortion value of each sampling point in the abnormal perturbation components with a preset polarization direction to construct a polarization-phase correlation matrix; forming a two-channel spatial field coupling gradient and polarization attributes; Taking the direction of the energy attenuation gradient vector field in the two-channel spatial field as the basic vector direction, adjusting the modulus of the basic vector by the polarization-phase correlation matrix in the two-channel spatial field according to the weight, and introducing the second-order derivative of the phase distortion space to correct the basic vector direction; performing directional enhancement filtering on the corrected basic vector: retaining the direction of potential damage propagation of the joint to form a three-dimensional damage-sensitive vector field.

[0015] Optionally, the generation process of the guided wave interference pattern includes the following steps: Applying orthogonal polarization in a specific area of the longitudinal joint, including a composite ultrasonic energy field of longitudinal wave and shear wave energy fluxes with a specific frequency combination; forming a spatio-temporal coupled guided wave interference pattern in the entire longitudinal joint area including the joint body and the concrete on both sides. The guided wave interference pattern includes the amplitude distribution, phase delay, and mode conversion characteristics under the interaction of the composite ultrasonic energy field.

[0016] Through the multi-modal guided wave interference excited by the composite ultrasonic energy field across the entire seam region of the present invention, coupling the amplitude-phase-mode conversion characteristics, it breaks through the dimensional limitations of traditional single-modal detection and realizes a three-dimensional sensitive response to hidden damages; the multi-physical field inverse decoupling technology extracts damage feature components from complex interference patterns, and the constructed three-dimensional vector field can simultaneously analyze the spatial distribution pattern of damages, the mechanical response directionality, and the material degradation trend. Based on the manifold topology analysis of the vector field, a mapping relationship between damage features and material property degradation is established. By identifying the damage core at the energy flow convergence point, determining the propagation front at the divergence point, and characterizing the performance degradation zone at the curl anomaly region, a complete topological description system of damage evolution is formed; the migration path tracking technology reveals the dynamic law of damage propagation and realizes the leap from static detection to dynamic prediction. The orthogonal polarization excitation enhances the sensitivity to micro-defects, the vector field construction realizes the identification of mesoscopic-scale damage patterns, and the topological analysis completes the prediction of macroscopic evolution trends, forming a cross-scale diagnostic chain; it effectively distinguishes the baseline features caused by seam geometric discontinuities from real damage perturbations and significantly improves the detection ability for early hidden damages. The time-space coupled guided wave interference mechanism suppresses the anisotropy interference of materials through mode conversion characteristics, and the multi-physical field decoupling algorithm automatically compensates for environmental noise, enabling the system to maintain stable diagnostic accuracy under complex working conditions; the entire set of technical systems realizes a closed-loop diagnosis from damage location, pattern recognition to condition assessment.

[0017] Other features and advantages of the present invention will be described in the following specification, and in part, will be obvious from the specification, or can be understood by implementing the present invention. The objectives and other advantages of the present invention can be achieved and obtained by the structures specifically pointed out in the written specification and the drawings.

[0018] The technical solutions of the present invention will be further described in detail below through the drawings and embodiments. Description of the Drawings

[0019] The drawings are used to provide a further understanding of the present invention and constitute a part of the specification. They are used together with the embodiments of the present invention to explain the present invention and do not constitute a limitation to the present invention. In the drawings: Figure 1 is the flow chart of the method for diagnosing hidden damages of longitudinal joint of shield tunnel based on multi-modal ultrasonic guided wave in Embodiment 1 of the present invention; Figure 2 is the schematic diagram of the method for diagnosing hidden damages of longitudinal joint of shield tunnel based on multi-modal ultrasonic guided wave in Embodiment 1 of the present invention; Figure 3 is the process diagram of forming a time-space coupled guided wave interference pattern in Embodiment 2 of the present invention; Figure 4 is the process diagram of constructing a three-dimensional damage-sensitive vector field inside the longitudinal joint in Embodiment 3 of the present invention; Figure 5 This is a process diagram for determining the dominant mode of hidden damage and the critical state level of its development in Embodiment 5 of the present invention. Specific Embodiment

[0020] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are only for the purpose of illustrating and explaining the present invention, and are not intended to limit the present invention.

[0021] The terms used in the embodiments of the present application are only for the purpose of describing specific embodiments, and are not intended to limit the embodiments of the present application. In the embodiments of the present application, the singular forms "a", "the", and "said" are also intended to include the 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 of the associated listed items.

[0022] 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 do not have to be 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 specific circumstances.

[0023] Embodiment 1: As Figure 1 shown, the embodiment of the present invention provides a method for diagnosing hidden damage of longitudinal joint of shield tunnel based on multi-modal ultrasonic guided wave, which includes the following steps: S100: Apply orthogonal polarization in a specific area of the longitudinal joint, including a composite ultrasonic energy field of longitudinal wave and transverse wave energy fluxes with a specific frequency combination; form a spatio-temporal coupled guided wave interference pattern in the whole area of the longitudinal joint including the joint body and the concrete on both sides, and the guided wave interference pattern includes the amplitude distribution, phase delay and mode conversion characteristics under the interaction of the composite ultrasonic energy field; S200: Use the obtained guided wave interference pattern, through multi-physical field reverse decoupling, separate the baseline interference characteristics generated by the geometric discontinuity of the joint in the guided wave interference pattern, and the abnormal disturbance components caused by hidden damage. The abnormal disturbance components are manifested as abnormal attenuation of energy and phase distortion in a specific polarization direction; form the spatial distribution and intensity of the abnormal disturbance components reverse decoupled by the multi-physical field, construct a three-dimensional damage-sensitive vector field inside the longitudinal joint, and quantify the response intensity and directionality of different positions to the damage mode; S300: Using the generated three-dimensional damage-sensitive vector field, perform dynamic manifold topology analysis to identify the topological features of the energy flow convergence points containing potential damage cores, the divergence points of the damage propagation frontiers, and the vector curl anomaly regions of the material property degradation zones in the three-dimensional damage-sensitive vector field, and track the migration paths and intensity evolution trajectories of the topological features with changes in the detection conditions; based on the migration path stability and intensity evolution rate of the topological features, determine the dominant mode of the hidden damage and the critical state level at which its development is located.

[0024] The working principle and beneficial effects of the above technical solution are as follows: In this embodiment, orthogonal polarization is first applied to a specific area of the longitudinal joint, and a composite ultrasonic energy field containing longitudinal and transverse wave energy fluxes with a specific frequency combination is formed; a spatio-temporal coupled guided wave interference pattern is formed in the entire area of the longitudinal joint including the joint body and the concrete on both sides, and the guided wave interference pattern includes the amplitude distribution, phase delay, and mode conversion characteristics under the interaction of the composite ultrasonic energy field; secondly, using the obtained guided wave interference pattern, through multi-physical field inverse decoupling, the baseline interference characteristics generated by the geometric discontinuity of the joint are separated. The longitudinal joint is a continuous structural segment formed by connecting adjacent segments through bolts, sealing materials, etc., and undertakes comprehensive functions such as load transfer and waterproofing; the joint body specifically refers to the contact interface (including filling materials, embedded parts, etc.) between the concrete segments inside the joint, which is the main area of stress concentration and damage initiation), and the abnormal disturbance component caused by hidden damage, and the abnormal disturbance component is manifested as abnormal energy attenuation and phase distortion in a specific polarization direction; based on the spatial distribution and intensity of the abnormal disturbance component obtained by multi-physical field inverse decoupling, a three-dimensional damage-sensitive vector field inside the longitudinal joint is constructed to quantify the response intensity and directionality of different positions to the damage mode; finally, using the generated three-dimensional damage-sensitive vector field, perform dynamic manifold topology analysis to identify the topological features of the energy flow convergence points containing potential damage cores, the divergence points of the damage propagation frontiers, and the vector curl anomaly regions of the material property degradation zones in the three-dimensional damage-sensitive vector field, and track the migration paths and intensity evolution trajectories of the topological features with changes in the detection conditions; based on the migration path stability and intensity evolution rate of the topological features, determine the dominant mode of the hidden damage and the critical state level at which its development is located (the principle is referred to in the appendix Figure 2). Through the synergistic effect of multi-modal excitation of orthogonal polarization composite ultrasonic energy field, inverse decoupling of multi-physical fields and dynamic manifold topology analysis, the full-dimensional diagnosis and evolution prediction of hidden damage in longitudinal joint of shield tunnel are realized. Its comprehensive technical effects are as follows: through the multi-modal guided wave interference excited by the composite ultrasonic energy field in the whole joint area, coupling the amplitude-phase-mode conversion characteristics, breaking through the dimensional limitation of traditional single-modal detection, and realizing the three-dimensional sensitive response to hidden damage; the multi-physical field inverse decoupling technology extracts the damage feature components from the complex interference pattern, and the constructed three-dimensional vector field can simultaneously analyze the spatial distribution pattern of damage, the mechanical response directionality and the material degradation trend. Based on the manifold topology analysis of the vector field, the mapping relationship between damage characteristics and material property degradation is established. By identifying the damage core at the energy flow convergence point, determining the expansion front at the divergence point, and characterizing the performance degradation zone at the curl anomaly area, a complete topological description system of damage evolution is formed; the migration path tracking technology reveals the dynamic law of damage expansion, realizing the leap from static detection to dynamic prediction. The orthogonal polarization excitation enhances the sensitivity to micro-defects, the vector field construction realizes the mesoscopic scale damage mode recognition, and the topological analysis completes the macroscopic evolution trend prediction, forming a cross-scale diagnosis chain; effectively distinguishing the baseline characteristics caused by joint geometric discontinuity from the real damage perturbation, significantly improving the detection ability of early hidden damage. The spatio-temporal coupled guided wave interference mechanism suppresses the material anisotropy interference through the mode conversion characteristics, and the multi-physical field decoupling algorithm automatically compensates the environmental noise, enabling the system to maintain stable diagnostic accuracy under complex working conditions; the whole set of technical systems realizes the closed-loop diagnosis from damage location, mode recognition to state assessment.

[0025] Embodiment 2: As Figure 3 shown, on the basis of Embodiment 1, the process of forming a spatio-temporal coupled guided wave interference pattern provided by the embodiment of the present invention includes the following steps: S101: Apply an orthogonal polarization composite ultrasonic energy field to a specific area of the longitudinal joint head, and penetrate the joint interface in a non-contact manner; S102: When the longitudinal wave energy flux of a specific frequency combination of the composite ultrasonic energy field propagates along the joint length direction, it repeatedly acts on the joint geometric interface to generate a periodic phase modulation wave train; the shear wave energy flux of the composite ultrasonic energy field forms a polarization-constrained shear waveguide in the concrete on both sides of the joint, and its energy flux direction is dynamically deflected by material anisotropy, forming a dynamic propagation network of coupled wave groups in the whole area of the longitudinal joint, including the real-time energy exchange between the longitudinal wave phase modulation wave train and the transverse wave polarization-constrained waveguide. S103: The coupled wave group dynamic propagation network is triggered at the junction area between the joint body and the concrete. The energy flux vectors of the longitudinal wave phase modulation wave train and the transverse wave polarization constraint waveguide are superposed, generating a spatially non-uniform distribution of extreme points of interference energy density. The transverse wave polarization plane is scattered by the microscopic structure of the concrete, causing the accumulation of phase lag related to the polarization direction. The energy fluxes of the longitudinal and transverse waves undergo asymmetric mode conversion in the damaged area, forming a spatio-temporal coupled guided wave interference pattern.

[0026] The working principle and beneficial effects of the above technical solution are as follows: In this embodiment, first, an orthogonally polarized composite ultrasonic energy field is applied to a specific area at the longitudinal joint head to penetrate the joint interface in a non-contact manner. Secondly, when the longitudinal wave energy flux of a specific frequency combination in the composite ultrasonic energy field propagates along the joint length direction, it repeatedly interacts with the joint geometric interface to generate a periodic phase modulation wave train. The transverse wave energy flux in the composite ultrasonic energy field forms a polarization-constrained shear waveguide in the concrete on both sides of the joint, and its energy flux direction is dynamically deflected by material anisotropy, forming a coupled wave group dynamic propagation network throughout the longitudinal joint, including the real-time energy exchange between the longitudinal wave phase modulation wave train and the transverse wave polarization constraint waveguide. Finally, the coupled wave group dynamic propagation network is triggered at the junction area between the joint body and the concrete. The energy flux vectors of the longitudinal wave phase modulation wave train and the transverse wave polarization constraint waveguide are superposed, generating a spatially non-uniform distribution of extreme points of interference energy density. The transverse wave polarization plane is scattered by the microscopic structure of the concrete, causing the accumulation of phase lag related to the polarization direction. The energy fluxes of the longitudinal and transverse waves undergo asymmetric mode conversion in the damaged area, forming a spatio-temporal coupled guided wave interference pattern. Through the multimodal coupling effect of the orthogonally polarized composite ultrasonic energy field, the above solution constructs a dynamic wave group propagation network in the longitudinal joint area, realizing the spatio-temporal precise regulation of the internal energy distribution of the structure. The orthogonally polarized longitudinal / transverse wave composite energy field penetrates the interface through non-contact excitation. The longitudinal wave energy flux forms a phase modulation wave train along the joint length direction, while the transverse wave energy flux establishes a polarization-constrained waveguide in the concrete. The two achieve dynamic energy exchange through material anisotropy, forming a coupled wave group network covering the entire joint area. The wave group network triggers the vector superposition of the longitudinal / transverse wave energy fluxes at the joint interface, generating a spatially non-uniform interference extreme point distribution. The transverse wave polarization plane is scattered by the microscopic structure to produce phase lag accumulation, and the asymmetric mode conversion in the damaged area further strengthens the spatio-temporal coupling characteristics of the interference pattern. The coupled wave group network maps the joint geometric characteristics, material anisotropy, and damage state into the spatio-temporal modulation signal of the interference pattern through the real-time energy exchange between the phase modulation wave train and the polarization constraint waveguide, realizing the cross-scale sensitive characterization of internal defects in the structure.

[0027] In summary, this embodiment forms a guided wave interference pattern with spatio-temporal resolution ability, providing a composite diagnostic basis for the detection of hidden defects in the joint structure by coupling the dynamic propagation field and the static interference field.

[0028] Example 3: As Figure 4As shown, on the basis of Embodiment 1, the process of constructing a three-dimensional damage-sensitive vector field inside a longitudinal joint provided by an embodiment of the present invention includes the following steps: S201: Discretely sample the polarization directions of the abnormal disturbance components output by the inverse decoupling of multiple physical fields over the entire domain of the longitudinal joint according to a preset spatial grid. Each sampling point captures the disturbance intensity and phase shift angle within a specific polarization plane; S202: Perform three-dimensional spatial gradient operations on the energy attenuation intensity of each sampling point in the abnormal disturbance components to generate an energy attenuation gradient vector field. The direction points to the direction of the maximum rate of change of energy attenuation, and the modulus represents the rate of change; Correlate the phase distortion value of each sampling point in the abnormal disturbance components with a preset polarization direction to construct a polarization-phase correlation matrix. The matrix elements quantify the contribution weight of a specific polarization direction to the phase distortion; Form a two-channel spatial field that couples gradients and polarization attributes, including the energy attenuation gradient vector field and the polarization-phase correlation matrix; S203: Use the direction of the energy attenuation gradient vector field in the two-channel spatial field as the basic vector direction. Adjust the modulus of the basic vector according to the weight of the polarization-phase correlation matrix in the two-channel spatial field, and introduce the second-order spatial derivative of the phase distortion space to correct the basic vector direction. The second-order derivative is calculated from the spatial distribution of the phase distortion value; Perform directional enhancement filtering on the corrected basic vector: Retain the direction of potential damage propagation of the joint to form a three-dimensional damage-sensitive vector field. The vector direction of each spatial point represents the damage response sensitive direction, and the vector modulus represents the response intensity of this position to damage.

[0029] The working principle and beneficial effects of the above technical solution are as follows: In this embodiment, first, the abnormal disturbance components output by the multi-physical field inverse decoupling are discretely sampled in terms of polarization direction according to a preset spatial grid in the entire longitudinal joint area, and the disturbance intensity and phase offset angle in a specific polarization plane are captured at each sampling point; secondly, a three-dimensional spatial gradient operation is performed on the energy attenuation intensity of each sampling point in the abnormal disturbance components to generate an energy attenuation gradient vector field, where the direction points to the direction of the maximum change rate of energy attenuation, and the modulus represents the change rate; the phase distortion value of each sampling point in the abnormal disturbance components is correlated with the preset polarization direction to construct a polarization-phase correlation matrix, and the matrix elements quantify the contribution weight of a specific polarization direction to the phase distortion; a dual-channel spatial field combining the coupling gradient and polarization attributes is formed, including the energy attenuation gradient vector field and the 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 modulus of the basic vector is adjusted by weight according to the polarization-phase correlation matrix in the dual-channel spatial field, and the direction of the basic vector is corrected by introducing the second-order spatial derivative of the phase distortion space, and the second-order derivative is calculated from the spatial distribution of the phase distortion value; directional enhancement filtering is performed on the corrected basic vector: retain the direction related to the potential damage propagation of the joint to form a three-dimensional damage-sensitive vector field, where the vector direction of each spatial point represents the damage response sensitive direction, and the vector modulus represents the response intensity of this position to the damage. The technical process of constructing the three-dimensional damage-sensitive vector field inside the longitudinal joint in the above solution 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 operation. By discretely sampling in terms of polarization direction, the energy attenuation gradient and the polarization characteristics of phase distortion of the abnormal disturbance are captured simultaneously, expanding the traditional single-field quantity detection to the energy-phase dual-channel coupling analysis; the energy attenuation gradient vector field reflects the spatial distribution law of material stiffness degradation, while the polarization-phase correlation matrix reveals the wave mode coupling effect caused by microscopic defects. The two form a holographic characterization of damage sensitivity through vector field operations. Based on the vector direction correction mechanism of the second-order spatial derivative of the phase distortion, the defect that the traditional gradient field is insufficiently sensitive to the damage propagation direction is overcome; the directional enhancement filtering algorithm combines the joint geometric constraint conditions, so that the finally generated vector field can not only reflect the current damage state, but also predict the potential crack propagation path through the vector topological structure. In the finally generated three-dimensional vector field, the vector parameters (direction angle and modulus) of each spatial point strictly correspond to the physical response mechanism of material damage: the vector direction represents the maximum damage-sensitive azimuth, and the modulus quantifies the degree of local damage accumulation; this diagnostic method based on the joint analysis of the polarization characteristics and energy dissipation of the wave field significantly improves the detection rate and localization accuracy of early micro-damage.

[0030] In summary, in this embodiment, by constructing a vector space mapping covering the disturbance characteristics of the entire frequency band, the implicit material damage information is converted into an explicit three-dimensional computable field, providing a decision-making basis with clear physical significance for structural health monitoring.

[0031] Embodiment 4: On the basis of Embodiment 3, the process of generating an energy attenuation gradient vector field provided by the embodiment of the present invention includes the following steps: S2021: Input the discrete sampling results of the polarization direction, extract the scalar values of the energy attenuation intensity of each spatial grid point, and form a three-dimensional discrete field of energy attenuation intensity covering the entire joint area. Each node of the three-dimensional discrete field of energy attenuation intensity stores the energy loss rate within a specific polarization plane at that position; S2022: Taking each discrete point in the three-dimensional discrete field of energy attenuation intensity as the center, dynamically determine its differential neighborhood range according to the seam geometric topology constraints; adopt a large-scale neighborhood along the seam length direction to capture the long-range attenuation trend; adopt a small-scale neighborhood perpendicular to the seam interface direction to capture the interface mutation characteristics; within the adaptive neighborhood, obtain the energy attenuation intensity difference between the center point and each neighborhood point, and generate a set of spatial intensity differences; form a set of spatial intensity gradient characteristics under geometric constraints, including the adaptive neighborhood range of each discrete point and its corresponding set of intensity differences; S2023: Perform a principal direction screening on the set of intensity differences of each discrete point in the set of spatial intensity gradient characteristics, retain the neighborhood points with the absolute value of the difference greater than the preset threshold, and eliminate noise interference; based on the screened neighborhood points, obtain the weighted vector sum of the intensity differences, where the weight is jointly determined by the reciprocal of the spatial distance between the neighborhood point and the center point and the seam orientation consistency coefficient, and the direction of the vector sum is the direction of the maximum change rate of energy attenuation, and the magnitude of the vector sum is normalized to a change rate quantization value; form an energy attenuation gradient vector field, where each discrete point corresponds to a vector, its direction represents the fastest increasing path of energy attenuation, and the magnitude represents the severity of attenuation.

[0032] The working principle and beneficial effects of the above technical solution are as follows: In this embodiment, first, the discrete sampling results of the polarization direction are input, and the scalar value of the energy attenuation intensity at each spatial grid point is extracted to form a three-dimensional discrete field of energy attenuation intensity covering the entire joint area. Each node of the three-dimensional discrete field of energy attenuation intensity stores the energy loss rate within a specific polarization plane at that position. Secondly, taking each discrete point in the three-dimensional discrete field of energy attenuation intensity as the center, its differential neighborhood range is dynamically determined according to the seam geometric topology constraints. A large-scale neighborhood is used along the seam length direction to capture the long-range attenuation trend, and a small-scale neighborhood is used perpendicular to the seam interface direction to capture the interface mutation characteristics. Within the adaptive neighborhood, the difference in energy attenuation intensity between the center point and each neighborhood point is obtained to generate a set of spatial intensity differences. A set of spatial intensity gradient features under geometric constraints is formed, including the adaptive neighborhood range of each discrete point and its corresponding set of intensity differences. Finally, the main direction of the set of intensity differences at each discrete point in the set of spatial intensity gradient features is screened, and the neighborhood points with the absolute value of the difference greater than the preset threshold are retained to eliminate noise interference. Based on the screened neighborhood points, the weighted vector sum of the intensity differences is obtained, and the weight is jointly determined by the reciprocal of the spatial distance between the neighborhood point and the center point and the seam orientation consistency coefficient. The direction of the vector sum is the direction of the maximum change rate of energy attenuation, and the magnitude of the vector sum is normalized to a change rate quantization value. An energy attenuation gradient vector field is formed, with each discrete point corresponding to a vector, whose direction represents the path of the fastest energy attenuation enhancement, and the magnitude represents the degree of attenuation severity. The above solution realizes the global quantitative characterization of the energy attenuation characteristics of the welded joint and the visual reconstruction of the spatial evolution law through a gradient field construction method under multi-scale geometric constraints. Geometrically adaptive gradient field modeling: Through the dynamic neighborhood range determination mechanism, a macroscopic-scale neighborhood is used along the seam length direction to capture the long-range energy transfer law, and a microscopic-scale neighborhood is used in the interface normal direction to analyze the interface mutation behavior, realizing the cross-scale coupling characterization of millimeter-scale structural features and micron-scale interface effects. Noise-robust gradient extraction: Based on the intensity difference weighting algorithm with main direction screening, through the spatial correlation analysis of the energy loss rate in the polarization plane, the interference of measurement noise on the gradient field direction is effectively suppressed, ensuring the topological continuity of the attenuation path recognition. Physically driven vector field construction: Combining the two-parameter weighting model of the reciprocal distance weight and the orientation consistency coefficient, the generated gradient vector simultaneously satisfies: ① the spatial distance attenuation principle ② the seam geometric constraint conditions, accurately reflecting the spatial distribution law of the energy attenuation rate and direction. Visual mapping of three-dimensional attenuation behavior: The process of converting the discrete scalar field into a continuous vector field realizes the complete mathematical description of the spatial gradient of the energy loss rate. Its vector direction corresponds to the direction of the maximum attenuation rate, and the magnitude represents the local energy dissipation intensity, providing a full-field quantitative basis for the reliability assessment of the joint.

[0033] In summary, the present embodiment forms an energy attenuation kinetic characterization system with clear physical meanings, which can support the analysis requirements of three-dimensional attenuation behaviors in engineering application scenarios such as welding defect identification and life prediction.

[0034] Embodiment 5: As Figure 5 shown, based on Embodiment 1, the process of determining the dominant mode of hidden damage and the critical state level at which it develops provided by the embodiment of the present invention includes the following steps: S301: Input the generated three-dimensional damage-sensitive vector field, and obtain the time-series data of the vector field in the same joint area under time-varying detection conditions with preset different mechanical loads and temperature and humidity states to form a dynamic vector field sequence. Each time slice contains the vector distribution in the whole space; S302: Analyze the local manifold structure of the vector field in each time slice of the three-dimensional damage-sensitive vector field, and integrate the positions of three types of features, namely, energy flow convergence points, divergence points, and vector curl anomaly regions, and the corresponding vector modulus length mean values and direction attributes into a spatio-temporal topological feature set; form a state matrix of time-varying topological features. The rows of the state matrix record different time points, and the columns record the spatial coordinates, intensities, direction angles, and regional sizes of the three types of features; S303: Analyze the convergence of the intensity evolution trajectory of the spatial coordinate sequence of each type of feature in the state matrix, and calculate the standard deviation of the included angle of 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 amount per unit time; form a multi-parameter evolution state diagnosis of hidden damage, including the dominant damage mode and the critical level.

[0035] The working principle and beneficial effects of the above technical solution are as follows: In this embodiment, first, the generated three-dimensional damage-sensitive vector field is input, and the time-series data of the vector field in the same joint area is obtained under the time-varying detection conditions of preset different mechanical loads and temperature and humidity states to form a dynamic vector field sequence. Each time slice contains the vector distribution in the whole space. Secondly, the local manifold structure of the vector field in each time slice of the three-dimensional damage-sensitive vector field is analyzed, and the positions of three types of features, namely the energy flow convergence point, divergence point, and vector curl anomaly area, and the corresponding mean vector modulus length and direction attributes are integrated into a spatio-temporal topological feature set. A state matrix of 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. Finally, the convergence analysis of the intensity evolution trajectory of the spatial coordinate sequence of each type of feature in the state matrix is carried out, and the standard deviation of the included angle of the displacement vectors at adjacent time points is calculated; the time-domain gradient integration is carried out on the intensity sequence of each type of feature in the state matrix to obtain the intensity change amount per unit time; a multi-parameter evolution state diagnosis of hidden damage is formed, including the dominant damage mode and the critical level. The above solution realizes the multi-dimensional dynamic diagnosis of the hidden damage mode of the welded structure and the quantitative evaluation of the critical state through a spatio-temporal coupled damage evolution analysis framework. Spatio-temporal correlation modeling of damage-sensitive features: Based on the extraction of time-varying topological features from the dynamic vector field sequence, the spatial distribution and time-series evolution of three types of damage-sensitive features, namely the energy flow convergence point, divergence point, and vector curl anomaly area, are correlated and mapped to construct a whole-process characterization system for damage initiation, propagation, and critical failure. Automatic identification of the dominant damage mode: By analyzing the local manifold structure and constructing the time-varying state matrix, the spatial distribution law of features corresponding to different damage modes (such as microcrack initiation, interface debonding, fatigue accumulation) is quantified, and unsupervised classification of damage types is realized. Dynamic prediction of the critical state: Combining the convergence analysis of the intensity evolution trajectory and the time-domain gradient integration, the non-linear coupling relationship between the spatial migration rate of damage features and the intensity evolution trend is revealed. Through the joint criterion of the standard deviation of the displacement vector included angle and the intensity change amount, the state grading of damage from stable propagation (level I) to accelerated propagation (level II) until critical instability (level III) is realized; Damage assessment with multi-physical field coupling: By comparing the time series of vector fields under the preset multi-field loading conditions of mechanics-thermodynamics-humidity, the evolutionary contributions of environmental factors and real damage are separated, and the environmental robustness of hidden damage diagnosis is improved.

[0036] In summary, this embodiment has established the full-chain monitoring ability from microscopic damage initiation to macroscopic failure, providing a dynamic evolution basis for the active early warning of the structural health state and the prediction of the remaining life.

[0037] Embodiment 6: On the basis of Embodiment 5, the process of forming the state matrix of time-varying topological features provided by the embodiment of the present invention includes the following steps: S3021: Input the generated dynamic vector field sequence. Within each time slice, divide the joint global domain into overlapping local manifold units, where the unit size is constrained by the seam geometric features to ensure coverage of the minimum damage scale; S3022: Perform energy flow convergence point detection, divergence point detection, and vector curl anomaly region detection on each local manifold unit; bind the spatial positions, intensity indices, and dominant directions of the three types of features into feature tuples; form a raw feature set with time-space-intensity coupling; Energy flow convergence point detection: When all vectors within a local manifold unit satisfy the condition that their directions point to a certain internal focus and their magnitudes decrease with the distance from the focus, record the focus coordinates and the mean vector magnitude within the unit; Divergence point detection: When the center point of a local manifold unit satisfies the condition that the vector points away from the center and the magnitude along the seam axis increases with the distance, record the center point coordinates and the axial magnitude gradient value; Vector curl anomaly region detection: Calculate the maximum angle mutation value between adjacent vector directions within a local manifold unit. When it exceeds the threshold, record the spatial range of the unit and the cumulative intensity of the angle mutation; The intensity index includes for the convergence point: mean magnitude; for the divergence point: magnitude gradient value; for the curl region: cumulative intensity of the angle mutation; The dominant direction includes the axial direction of the divergence point / the maximum mutation direction of the curl region; Each feature tuple in the raw feature set with time-space-intensity coupling includes a timestamp, spatial coordinates, intensity index, and direction attribute; S3023: Classify and aggregate all features in the raw feature set with time-space-intensity coupling according to the timestamp and feature type; perform spatial clustering and merging on multiple instances of each type of feature at the same moment; for the merged features, calculate their spatial centroid coordinates, the mean value of the merged intensity index, and the synthesized value of the dominant direction; fill the matrix rows with the merged features at each time point according to a fixed dimension; form a state matrix of time-varying topological features; Among them, for the convergence point / divergence point in spatial clustering and merging: merge adjacent points with an Euclidean distance less than the seam thickness; for the curl anomaly region: merge regions with overlapping boundaries; the synthesized value of the dominant direction is the vector average; the fixed dimension includes spatial coordinate x3, intensity x1, direction angle x3, and region size x1; the rows of the state matrix = time points, the columns = feature parameters, and the matrix elements are the numerical expressions of three-dimensional spatial coordinates / intensity / direction / size.

[0038] The working principle and beneficial effects of the above technical solution are as follows: In this embodiment, a sequence of generated dynamic vector fields is first input. Within each time slice, the joint global domain is divided into overlapping local manifold units, and the unit size is constrained by the seam geometric features to ensure coverage of the minimum damage scale. Secondly, energy flow convergence point detection, divergence point detection, and vector curl anomaly region detection are performed on each local manifold unit; the spatial positions, intensity indices, and dominant directions of the three types of features are bound into feature tuples; an original feature set with time-space-intensity coupling is formed. Finally, all features in the original feature set with time-space-intensity coupling are classified and aggregated according to the time stamp and feature type; spatial clustering and merging are performed on multiple instances of each type of feature at the same moment; for the merged features, their spatial centroid coordinates, mean value of the merged intensity index, and synthesized value of the dominant direction are calculated; the merged features at each time point are filled into the matrix rows according to a fixed dimension; a state matrix of time-varying topological features is formed. The above solution constructs a structured characterization system for the evolution of hidden damage in welded joints through multi-scale feature fusion and space-time aggregation of dynamic vector fields. Through the division of local manifold units constrained by seam geometric features, the scale adaptation between macroscopic structural features and microscopic damage signals is realized; based on the collaborative detection of energy flow convergence points, divergence points, and vector curl anomaly regions, three typical behavior patterns of energy dissipation, stress concentration, and material heterogeneity during the damage evolution process are completely captured. By using the construction method of the original feature set with time-space-intensity coupling, the spatial distribution, intensity evolution, and direction evolution of the three types of damage features are uniformly parameterized, forming a four-dimensional feature tuple including time stamp, spatial coordinates, intensity index, and direction attribute, providing a multi-dimensional quantitative benchmark for damage dynamic monitoring. Through the spatial clustering and merging algorithm based on the seam thickness, the intelligent fusion of multi-source damage features is realized: the Euclidean distance constraint is used for the merging of discrete convergence / divergence points, and the boundary overlap detection is used for the merging of continuously distributed curl anomaly regions; this process not only retains the true physical size of the damaged area but also eliminates detection redundancy. The finally generated state matrix is transformed into a computable data structure through parameterized expression with a fixed dimension; each element in the matrix corresponds to the numerical expression of the spatial coordinates, intensity, direction, and regional size of a certain type of damage feature at a specific time point, providing a standardized input for subsequent damage pattern recognition and critical state assessment.

[0039] In summary, this embodiment establishes a complete conversion chain from the original vector field to the structured state matrix, realizes the accurate mapping of hidden damage features from the physical space to the data space, and provides core data support for the quantitative diagnosis of the health status of welded structures.

[0040] Embodiment 7: On the basis of Embodiment 5, the process of forming a multi-parameter evolution state diagnosis of hidden damage provided by the embodiment of the present invention includes the following steps: S3031: Input the generated state matrix, whose row vectors contain the spatial coordinate sequences, intensity sequences, and direction angle sequences of three types of features, namely energy flow convergence points, divergence points, and regions with abnormal vector curl, in terms of time series. S3032: For the spatial coordinate sequences of each type of feature, calculate the coordinate differences between adjacent time points to form a displacement vector sequence, perform a standard deviation operation on the direction angles of all adjacent displacement vectors in the spatial coordinate sequence to generate a path fluctuation coefficient; for the intensity sequences of each type of feature, calculate the intensity differences between adjacent time points, and perform a cumulative absolute value integration on the difference sequence along the time axis to generate an intensity transition cumulative quantity; form a two-channel parameter set that couples the spatial and intensity evolutions, with each type of feature corresponding to a path fluctuation coefficient and an intensity transition cumulative quantity. S3033: If the path fluctuation coefficient of the divergence point is greater than the first preset threshold and its intensity transition cumulative quantity is the largest, it is determined to be dominated by tensile debonding; if the intensity transition cumulative quantity of the region with abnormal vector curl and the path fluctuation coefficient of the energy flow convergence point increase synchronously, it is determined to be dominated by material degradation; judge the critical level according to the relationship between the intensity transition cumulative quantity and the preset threshold. Critical state mapping: When it is dominated by tensile debonding, if the intensity transition cumulative quantity is less than the second preset threshold, it is the stable latency period; if the intensity transition cumulative quantity is between the second preset threshold and the third preset threshold, including the second preset threshold and the third preset threshold, it is the subcritical expansion period; if the intensity transition cumulative quantity is greater than the third preset threshold, it is the near instability period. When it is dominated by material degradation: if the path fluctuation coefficient of the energy flow convergence point is less than the fourth preset threshold, it is the stable latency period; if the ratio of the parameters of the curl region to the convergence point is greater than the fifth preset threshold, it is the near instability period.

[0041] The working principle and beneficial effects of the above technical solution are as follows: In this embodiment, the generated state matrix is first input, and its row vectors include the spatial coordinate sequences, intensity sequences, and direction angle sequences of three types of features, namely energy flow convergence points, divergence points, and regions with abnormal vector curl, in the time series; secondly, for the spatial coordinate sequences of each type of feature, the coordinate differences between adjacent time points are calculated to form a displacement vector sequence, and the standard deviation operation is performed on the direction angles of all adjacent displacement vectors of the spatial coordinate sequence to generate a path fluctuation coefficient; for the intensity sequences of each type of feature, the intensity differences between adjacent time points are calculated, and the cumulative absolute value integral of the difference sequence along the time axis is performed to generate an intensity transition cumulative quantity; a two-channel parameter set that couples spatial and intensity evolution is formed, and each type of feature corresponds to a path fluctuation coefficient and an intensity transition cumulative quantity; finally, if the path fluctuation coefficient of the divergence point is greater than the first preset threshold and its intensity transition cumulative quantity is the largest, it is determined to be dominated by tensile debonding; if the intensity transition cumulative quantity of the region with abnormal vector curl and the path fluctuation coefficient of the energy flow convergence point increase synchronously, it is determined to be dominated by material degradation; according to the relationship between the intensity transition cumulative quantity and the preset threshold, the critical level is judged; critical state mapping: when it is dominated by tensile debonding, the intensity transition cumulative quantity is less than the second preset threshold, which is the stable latency period; the intensity transition cumulative quantity is between the second preset threshold and the third preset threshold, including the second preset threshold and the third preset threshold, which is the subcritical expansion period; the intensity transition cumulative quantity is greater than the third preset threshold, which is the near instability period; when it is dominated by material degradation: the path fluctuation coefficient of the energy flow convergence point is less than the fourth preset threshold, which is the stable latency period; the ratio of the parameters of the curl region to the convergence point is greater than the fifth preset threshold, which is the near instability period. The above solution realizes the cross-scale state mapping of the hidden damage of the longitudinal joint of the shield tunnel from microscopic evolution to macroscopic failure through multi-parameter dynamic coupling analysis. Decoupling of damage modes and identification of dominant mechanisms, through the two-channel parameter coupling of the path fluctuation coefficient and the intensity transition cumulative quantity, decomposes the complex multi-physical field perturbations at the joint interface into quantitative characterizations of two typical damage modes, namely tensile debonding (dominated by interface separation) and material degradation (dominated by deterioration of the body performance); the direction dispersion of the displacement vector sequence reflects the directional instability of damage propagation, and the time-domain cumulative characteristics of the intensity difference reveal the energy dissipation rate, and the two work together to overcome the defect of mode confusion in traditional single-parameter diagnosis. Dynamic calibration of the criticality of the evolution stage, based on the multi-level criterion system of parameter thresholds, constructs a state phase diagram of damage evolution: stable latency period (damage initiation but no through-path formation), subcritical expansion period (damage developing directionally but not exceeding the structural tolerance), near instability period (damage accumulation triggering non-linear failure); this grading mechanism realizes the precise capture from quantitative change to qualitative change through the non-linear growth characteristics of the intensity transition cumulative quantity and the spatial correlation of topological features (such as the ratio of the parameters of the curl region to the convergence point).The spatio-temporal coupling mechanism of failure warning, and the dynamic manifold analysis of spatio-temporal coordinate sequences establish the ternary topological correlation of the damage core (energy flow convergence point), the expansion front (divergence point), and the performance degradation zone (vector curl anomaly zone); 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 decision-making.

[0042] In summary, through the inversion of the constitutive relationship between the multi-modal response of guided waves and damage evolution in this embodiment, the implicit material deterioration process is transformed into a quantifiable dynamic parameter system, realizing the leap from the description of structural health status to the prediction of its mechanism. The dominant mode depends on the coupling relationship of three types of characteristic parameters. Tensile debonding is independently determined by the sudden increase of the divergence point parameter; material degradation is determined by the synchronization of the curl zone and the convergence point parameters; the critical level mapping parameter absolute value and relative relationship: the first preset threshold, the second preset threshold, the third preset threshold, and the fourth preset threshold are preset by the constitutive characteristics of the joint material; the parameter ratio (curl zone / convergence point) quantifies the correlation strength between material degradation and the damage core; there is no probability model throughout the process: the diagnostic logic is constructed based on the deterministic relationship of physical parameters.

[0043] Embodiment 8: On the basis of Embodiment 7, the process of generating the path fluctuation coefficient and the cumulative amount of generation intensity transition provided by the embodiment of the present invention includes the following steps: S30321: For the input state matrix row vector, extract the spatial coordinate time series chain of the three-dimensional coordinates of each type of feature at consecutive time points and the intensity value time series chain of the intensity index values at consecutive time points; form the original sequence of feature spatio-temporal evolution; S30322: For the spatial coordinate time series chain of each feature in the original sequence of feature spatio-temporal evolution, calculate the vector difference of the coordinates at adjacent time points to generate a displacement vector sequence, where each vector represents the spatial migration direction and distance of the feature within a unit time; for the intensity value time series chain of each feature in the original sequence of feature spatio-temporal evolution, calculate the algebraic difference of the intensity values at adjacent time points to generate an intensity difference sequence, with positive values indicating enhancement and negative values indicating weakening; form a two-channel evolution field that couples spatial movement and intensity change, where the displacement vector sequence describes the evolution path of the feature position and the intensity difference sequence describes the intensity transition behavior of the feature; Among them, based on the spatial coordinate time series chain in the original feature spatio-temporal evolution sequence, by subtracting the corresponding three-dimensional coordinate components at adjacent time points, subtracting the coordinates of the later time point from those of the previous time point, a displacement vector sequence describing the spatial migration direction and distance of the feature is generated; simultaneously, based on the intensity value time series chain, by directly subtracting the intensity values at adjacent time points, subtracting the intensity of the later time point from that of the previous time point, an intensity difference sequence reflecting the increasing / decreasing trend of the feature intensity is generated; for the time continuity feature, the displacement vector sequence provides a basis for direction analysis for the path fluctuation coefficient, and the intensity difference sequence provides a basis for quantifying the sudden change for the intensity jump cumulative amount. The two together constitute the core elements of the dual-channel evolution field; S30323: Extract the direction unit vectors of all adjacent displacement vectors from the displacement vector sequence of the dual-channel evolution field, calculate the arccosine value of the dot product of each pair of adjacent direction vectors, calculate the standard deviation of the arccosine values of the dot products, 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 jump cumulative amount; form a dual-channel evolution parameter set, and each feature corresponds to two parameters, namely the path fluctuation coefficient and the intensity jump cumulative amount.

[0044] The working principle and beneficial effects of the above technical solution are as follows: In this embodiment, first, 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 feature at consecutive time points and the intensity value time series chain of the intensity index values at consecutive time points; a raw sequence of feature spatio-temporal evolution is formed. Secondly, for the spatial coordinate time series chain of each feature in the raw sequence of feature spatio-temporal evolution, the vector difference between the coordinates at adjacent time points is calculated to generate a displacement vector sequence, and each vector represents the spatial migration direction and distance of the feature within a unit time. For the intensity value time series chain of each feature in the raw sequence of feature spatio-temporal evolution, the algebraic difference between the intensity values at adjacent time points is calculated to generate an intensity difference sequence, where a positive value indicates enhancement and a negative value indicates weakening. A dual-channel evolution field coupling spatial motion and intensity change is formed. The displacement vector sequence describes the evolution path of the feature position, and the intensity difference sequence describes the intensity transition behavior of the feature. Finally, the unit direction vectors of all adjacent displacement vectors are extracted from the displacement vector sequence of the dual-channel evolution field, the arccosine value of the dot product of each pair of adjacent direction vectors is calculated, and the standard deviation of the arccosine values of the dot products is calculated to output the path fluctuation coefficient; 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 cumulative amount; a set of dual-channel evolution parameters is formed, and each feature corresponds to two parameters, namely the path fluctuation coefficient and the intensity transition cumulative amount. The above solution realizes the quantitative characterization of dynamic feature evolution behavior through spatio-temporal coupling analysis; constructs a dual-channel evolution field to synchronously capture the spatio-temporal coupling relationship between the spatial displacement vector and the intensity difference sequence, and then extracts high-order feature parameters through vector operations and sequence analysis. The path fluctuation coefficient quantifies the frequency and amplitude of the direction mutation of the feature motion trajectory through the arccosine operation of the unit direction vectors of the displacement vector sequence, and its standard deviation represents the elimination of the dimension influence and accurately reflects the tortuosity and uncertainty of the path; this parameter overcomes the defect that the traditional Euclidean distance metric cannot capture direction changes. The intensity transition cumulative amount adopts the strategy of absolute value linear weighted accumulation, which not only retains the original polarity information of the intensity change (through the preprocessing of the difference sequence), but also highlights the total transition amount through the absolute value conversion, and the weighted accumulation strengthens the contribution degree of recent changes; this method effectively distinguishes different evolution modes such as continuous enhancement and intermittent fluctuation. The collaborative output of the dual-channel parameters constructs a complete spatio-temporal evolution feature space: the path fluctuation coefficient describes the geometric characteristics of the motion trajectory, and the intensity transition cumulative amount characterizes the energy transfer efficiency. The two maintain the consistency of physical meaning through the original sequence generation mechanism of spatio-temporal alignment; this coupled quantification method improves the analysis granularity of complex evolution behavior compared with single-dimensional time series analysis.

[0045] In summary, through the three-stage processing of spatio-temporal chain decomposition - vector differential operation - multi-channel statistical modeling in this embodiment, the computable conversion of the nonlinear evolution process is realized, providing a feature engineering framework with physical interpretability for scenarios such as the stability analysis and abnormal pattern detection of dynamic systems.

[0046] Embodiment 9: On the basis of Embodiment 8, the process of the output path fluctuation coefficient provided by the embodiment of the present invention includes the following steps: S303231: Input the generated displacement vector sequence, perform direction normalization processing on each displacement vector in the sequence to form a set of direction unit vectors of the displacement vector sequence, and each vector represents the pure migration direction of the feature within a unit time; S303232: For the time series chain of direction unit vectors in the set of direction unit vectors, take each pair of adjacent vectors in chronological order, calculate the dot product value of each pair of adjacent vectors, perform an inverse cosine function operation on the dot product value to generate a direction deflection angle sequence, and each angle value quantifies the instantaneous deflection amplitude of the migration direction in adjacent time periods; form a continuous direction deflection perturbation record, and the angle sequence directly maps the severity of the path direction change; S303233: Perform global statistical dispersion analysis on the direction deflection angle sequence in the continuous direction deflection perturbation record, calculate the arithmetic mean of all angles in the direction deflection angle sequence, calculate the sum of the squares of the absolute deviations of each angle from the mean value, and perform a root mean square operation on the sum of squares; form a path fluctuation coefficient. The larger the value of the path fluctuation coefficient, the stronger the disorder of the migration direction and the lower the path stability.

[0047] The working principle and beneficial effects of the above technical solution are as follows: In this embodiment, a generated displacement vector sequence is first input, and each displacement vector in the sequence is subjected to direction normalization processing to form a set of direction unit vectors of the displacement vector sequence. Each vector represents the pure direction of feature migration within a unit time. Secondly, for the time series chain of direction unit vectors in the set of direction unit vectors, every pair of adjacent vectors is taken in chronological order, and the dot product value of each pair of adjacent vectors is calculated. The inverse cosine function operation is performed on the dot product value to generate a direction deflection angle sequence. Each angle value quantifies the instantaneous deflection amplitude of the migration direction in adjacent time periods. A continuous direction deflection perturbation record is formed, and the angle sequence directly maps the severity of the path direction change. Finally, global statistical dispersion analysis is performed on the direction deflection angle sequence in the continuous direction deflection perturbation record. The arithmetic mean of all angles in the direction deflection angle sequence is calculated, the sum of the squares of the absolute deviations of each angle from the mean is calculated, and the root mean square operation is performed on the sum of squares. A path fluctuation coefficient is formed. The larger the value of the path fluctuation coefficient, the stronger the disorder of the migration direction and the lower the path stability. The above solution realizes the quantitative characterization of the direction stability of the motion trajectory through the integration of vector differential geometry and statistical dynamics methods. The multi-order analysis of the direction perturbation degree decomposes the continuous motion trajectory into an instantaneous direction deflection sequence through the normalization processing of the displacement vector and the inverse cosine operation of the dot product of adjacent vectors, overcoming the dependence of traditional curvature calculation on the assumption of a continuous path and directly capturing the direction mutation characteristics at the discrete time scale. Statistical modeling of path disorder: Based on the root mean square dispersion calculation of the direction deflection angle sequence, a fluctuation coefficient with rotational invariance is constructed. Its physical meaning is the energy density of direction perturbation. This coefficient amplifies the abnormal deflection events in non-stationary motion through the second-order statistic (root mean square of the sum of squared deviations), and has a strong discrimination ability for non-deterministic paths such as random walks and Brownian motions. Completeness of stability evaluation: The finally output path fluctuation coefficient integrates the instantaneous amplitude of direction change (inverse cosine operation) and long-term statistical characteristics (root mean square), which can not only reflect the local influence of a single severe turn, but also characterize the anisotropy degree of the overall motion pattern. This time-frequency joint analysis characteristic makes it superior to traditional methods based solely on curvature integration or Fourier transform.

[0048] In summary, in this embodiment, through the cascade of normalization, dot product, inverse cosine, and root mean square dispersion, the geometrically intuitive direction change is transformed into a stability index with a strict probability interpretation, providing differentiable high-order features for tasks such as motion trajectory classification and phase transition detection of dynamic systems.

[0049] Embodiment 10: On the basis of Embodiment 9, the steps for calculating the dot product value of each pair of adjacent vectors provided by the embodiment of the present invention include: S3032321: Input the time series chain of direction unit vectors, extract all combinations of vectors at consecutive moments in the detection time order, and generate pairs of adjacent direction vectors in time series; S3032322: This is executed for each adjacent vector pair in the temporally adjacent direction vector pairs. The three-dimensional components of the two vectors are multiplied correspondingly to generate a direction consistency scalar value, which characterizes the spatial coincidence degree of the directions of the two vectors; a direction consistency scalar sequence is formed, and each scalar value uniquely quantifies the strength of the continuity of the migration direction in adjacent time periods. S3032323: The scalar sequence in the direction consistency scalar sequence is used as the input parameter for calculating the direction deflection angle, which determines the output range of the arccosine operation; a dot product value sequence is formed.

[0050] The working principle and beneficial effects of the above technical solution are as follows: In this embodiment, first, a time-series chain of direction unit vectors is input, and all continuous moment vector combinations are extracted in the order of detection time to generate temporally adjacent direction vector pairs; secondly, for each adjacent vector pair in the temporally adjacent direction vector pairs, the three-dimensional components of the two vectors are multiplied correspondingly to generate a direction consistency scalar value, which characterizes the spatial coincidence degree of the directions of the two vectors; a direction consistency scalar sequence is formed, and each scalar value uniquely quantifies the strength of the continuity of the migration direction in adjacent time periods; finally, the scalar sequence in the direction consistency scalar sequence is used as the input parameter for calculating the direction deflection angle, which determines the output range of the arccosine operation; a dot product value sequence is formed; by multiplying the three-dimensional components of adjacent unit direction vectors correspondingly to obtain the dot product value, the direction consistency scalar sequence, and the scalar value is strictly limited within the interval [-1, 1], which directly corresponds to the cosine value of the spatial angle between the two vectors; when this sequence is input into the arccosine function, its value range characteristic naturally restricts the output angle range to be between [0, π] radians, thereby transforming the geometric direction continuity problem into a differentiable scalar operation chain, ensuring that the physical meaning of the direction deflection angle is clear and the calculation is stable. The above solution realizes the precise quantification of the continuity of the motion direction through the collaborative processing of vector algebraic operations and geometric transformations. Frame-by-frame analysis of direction consistency: Through the component-by-component dot product operation of temporally adjacent vector pairs, the direction change in three-dimensional space is transformed into a scalar sequence, which not only retains the spatial angle information between vectors but also achieves a balance between calculation efficiency and geometric meaning through scalarization; it overcomes the singularity problem when directly calculating Euler angles and is applicable to the analysis of motion trajectories in any dimension. Differentiable representation of dynamic continuity: The generated direction consistency scalar sequence is essentially a time-evolution record of direction cosine, and its value range (-1 to 1) strictly corresponds to the physical constraints of the vector angle. Negative values represent direction reversal, zero values represent orthogonal mutations, and positive values represent gradual turning; the normalized output provides a mathematically complete input space for subsequent arccosine operations. Structural guarantee of calculation stability: By forcing temporally adjacent constraints and deterministic operations of component-by-component multiplication, the common cumulative error problem in trajectory analysis is avoided; the scalar sequence, as an intermediate variable, not only isolates the original data noise but also maintains the reversibility of geometric transformations through linear operations, providing a numerically stable basis for high-order feature extraction.

[0051] In summary, in this embodiment, by establishing a cascaded data processing chain of vector pairs, scalar sequences, and arccosine inputs, the direction continuity detection in kinematics is transformed into a linear algebra problem with strict mathematical definitions, and its output can directly support stability analysis based on differential geometry or pattern classification based on statistical mechanics.

[0052] 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 equivalent technologies of the present invention, the present invention also intends to include these changes and modifications.

Claims

1. A method for diagnosing hidden damage of longitudinal joint of shield tunnel based on multimodal ultrasonic guided wave, characterized in that, It includes the following steps: Using the generated three-dimensional damage-sensitive vector field, perform dynamic manifold topology analysis to identify the topological features of the energy flow convergence points containing potential damage cores, the divergence points of the damage propagation frontiers, and the vector curl anomaly regions of the material property degradation zones in the three-dimensional damage-sensitive vector field, and track the migration paths and intensity evolution trajectories of the topological features with the changes in the detection conditions; According to the migration path stability and intensity evolution rate of the topological features, determine the dominant mode of the hidden damage and the critical state level at which its development is located.

2. The method for diagnosing hidden damage of longitudinal joint of shield tunnel based on multimodal ultrasonic guided wave according to claim 1, characterized in that The process of determining the dominant mode of the hidden damage and the critical state level at which its development is located includes the following steps: Input the generated three-dimensional damage-sensitive vector field, and obtain the time-series data of the vector field in the same joint area under the time-varying detection conditions of preset different mechanical loads and temperature and humidity states to form a dynamic vector field sequence, and each time slice contains the vector distribution of the entire space. Perform local manifold structure analysis on the vector field of each time slice in the three-dimensional damage-sensitive vector field, and integrate the positions of the three types of features of the energy flow convergence points, divergence points, and vector curl anomaly regions, as well as the corresponding vector modulus length mean and direction attributes into a spatio-temporal topological feature set; form a state matrix of the time-varying topological features, where the rows of the state matrix record different time points, and the columns record the spatial coordinates, intensities, direction angles, and regional sizes of the three types of features. Perform convergence analysis on the intensity evolution trajectories of the spatial coordinate sequences of each type of feature in the state matrix, and calculate the standard deviation of the included angles of the displacement vectors at adjacent time points; perform time-domain gradient integration on the intensity sequences of each type of feature in the state matrix to obtain the intensity change amount per unit time; form a multi-parameter evolution state diagnosis of the hidden damage, including the dominant damage mode and the critical level.

3. The method for diagnosing hidden damage of longitudinal joint of shield tunnel based on multimodal ultrasonic guided wave according to claim 2, characterized in that, The process of forming a state matrix of the time-varying topological features includes the following steps: Input the generated dynamic vector field sequence, and within each time slice, divide the entire joint area into overlapping local manifold units, and the unit size is constrained by the joint geometric features to ensure that the minimum damage scale is covered. Perform energy flow convergence point detection, divergence point detection, and vector curl anomaly region detection on each local manifold unit; bind the spatial positions, intensity indexes, and dominant directions of the three types of features into feature tuples; form a time-space-intensity coupled original feature set. Classify and aggregate all the features in the time-space-intensity coupled original feature set according to the time stamp and feature type. Perform spatial clustering and merging on multiple instances of each type of feature at the same moment; for the merged features, calculate their spatial centroid coordinates, the mean value of the merged intensity indexes, and the synthesized value of the dominant direction; fill the matrix rows with the merged features at each time point according to a fixed dimension; form a state matrix of the time-varying topological features.

4. The method for diagnosing hidden damage of longitudinal joint of shield tunnel based on multimodal ultrasonic guided wave according to claim 2, characterized in that, The process of forming a multi-parameter evolution state diagnosis of the hidden damage includes the following steps: Input the generated state matrix, and its row vectors include the spatial coordinate sequences, intensity sequences, and direction angle sequences of the three types of features of the energy flow convergence points, divergence points, and vector curl anomaly regions in time series. For the spatial coordinate sequence of each type of feature, calculate the coordinate differences at adjacent time points to form a displacement vector sequence, perform a standard deviation operation on the direction 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, calculate the intensity differences at adjacent time points, and perform a cumulative absolute value integration on the difference sequence along the time axis to generate an intensity transition cumulative quantity; Form a two-channel parameter set that couples spatial and intensity evolution, with each type of feature corresponding to a path fluctuation coefficient and an intensity transition cumulative quantity; If the path fluctuation coefficient of the divergence point is greater than the first preset threshold and its intensity transition cumulative quantity is the largest, it is determined to be dominated by tensile debonding; If the intensity transition cumulative quantity in the vector curl anomaly region and the path fluctuation coefficient of the energy flow convergence point increase synchronously, it is determined to be dominated by material degradation; based on the relationship between the intensity transition cumulative quantity and the preset threshold, judge the critical level.

5. The method for diagnosing hidden damage of longitudinal joint of shield tunnel based on multimodal ultrasonic guided wave according to claim 4, characterized in that, Critical state mapping: When it is dominated by tensile debonding, if the intensity transition cumulative quantity is less than the second preset threshold, it is a stable latency period; if the intensity transition cumulative quantity 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; if the intensity transition cumulative quantity is greater than the third preset threshold, it is a near instability period; When it is dominated by material degradation: if the path fluctuation coefficient of the energy flow convergence point is less than the fourth preset threshold, it is a stable latency period; if the ratio of the parameters in the curl region to the convergence point is greater than the fifth preset threshold, it is a near instability period.

6. The method for diagnosing hidden damage of longitudinal joint of shield tunnel based on multimodal ultrasonic guided wave according to claim 4, characterized in that, The process of generating the path fluctuation coefficient and generating the intensity transition cumulative quantity includes the following steps: For the row vector of the input state matrix, extract the spatial coordinate time series chain of the three-dimensional coordinates at consecutive time points of each type of feature and the intensity value time series chain of the intensity index values at consecutive time points; form the original sequence of feature spatio-temporal evolution; For the spatial coordinate time series chain of each feature in the original sequence of feature spatio-temporal evolution, calculate the vector difference of the coordinates at adjacent time points to generate a displacement vector sequence; for the intensity value time series chain of each feature in the original sequence of feature spatio-temporal evolution, calculate the algebraic difference of the intensity values at adjacent time points to generate an intensity difference sequence; form a two-channel evolution field that couples spatial motion and intensity change, where the displacement vector sequence describes the evolution path of the feature position, and the intensity difference sequence describes the intensity transition behavior of the feature; Extract the unit direction vectors of all adjacent displacement vectors from the displacement vector sequence in the two-channel evolution field, calculate the arccosine value of the dot product of each pair of adjacent direction vectors, calculate the standard deviation of the arccosine values of the dot products, and output the path fluctuation coefficient; take the absolute value of the intensity difference sequence in the two-channel evolution field, and perform a linear weighted accumulation on the absolute value sequence along the time axis to output the intensity transition cumulative quantity; form a two-channel evolution parameter set, with each feature corresponding to two parameters: the path fluctuation coefficient and the intensity transition cumulative quantity.

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

8. The method for diagnosing hidden damage of longitudinal joint of shield tunnel based on multimodal ultrasonic guided wave according to claim 1, wherein The generation process of the three-dimensional damage-sensitive vector field includes the following steps: Using the obtained guided wave interference pattern, through multi-physical field inverse decoupling, the baseline interference characteristics generated by the seam geometric discontinuity in the guided wave interference pattern and the abnormal perturbation components caused by hidden damage are separated. The abnormal perturbation components are manifested as the abnormal attenuation of energy and phase distortion in a specific polarization direction; forming the spatial distribution and intensity of the abnormal perturbation components decoupled by multi-physical field inverse, constructing the three-dimensional damage-sensitive vector field inside the longitudinal joint, and quantifying the response intensity and directionality of different positions to damage modes.

9. The method for diagnosing hidden damage of longitudinal joint of shield tunnel based on multimodal ultrasonic guided wave according to claim 8, wherein, The process of constructing the three-dimensional damage-sensitive vector field inside the longitudinal joint includes the following steps: Discretely sample the polarization direction of the abnormal perturbation components output by multi-physical field inverse decoupling according to the preset spatial grid in the entire longitudinal joint area, and each sampling point captures the perturbation intensity and phase offset angle in a specific polarization plane; Perform three-dimensional spatial gradient operation on the energy attenuation intensity of each sampling point in the abnormal perturbation components to generate an energy attenuation gradient vector field; Associate the phase distortion value of each sampling point in the abnormal perturbation components with the preset polarization direction to construct a polarization-phase correlation matrix; form a two-channel spatial field coupling gradient and polarization attributes; Use the direction of the energy attenuation gradient vector field in the two-channel spatial field as the basic vector direction, adjust the modulus of the basic vector by the polarization-phase correlation matrix in the two-channel spatial field according to the weight, and introduce the second-order derivative of the phase distortion space to correct the basic vector direction; perform directional enhancement filtering on the corrected basic vector: retain the direction of potential damage propagation of the seam to form a three-dimensional damage-sensitive vector field.

10. The method for diagnosing hidden damage of longitudinal joint of shield tunnel based on multimodal ultrasonic guided wave according to claim 9, wherein The generation process of the guided wave interference pattern includes the following steps: Apply orthogonal polarization in a specific area of the longitudinal joint, including a composite ultrasonic energy field of longitudinal and transverse wave energy fluxes with specific frequency combinations; form a spatio-temporal coupled guided wave interference pattern in the entire longitudinal joint area including the seam body and the concrete on both sides. The guided wave interference pattern includes the amplitude distribution, phase delay, and mode conversion characteristics under the interaction of the composite ultrasonic energy field.

Images