Targeted diagnosis and treatment method for underground structure cracking in hot and humid areas in combination with nondestructive testing

Abstract

The invention relates to the technical field of underground structure engineering repair, and discloses a targeted diagnosis and treatment method for underground structure cracking in a hot and humid area in combination with nondestructive testing. The method comprises the following steps: firstly, collecting multi-source signals such as three-dimensional deformation data and crack distribution parameters of an underground structure, and constructing a structural damage feature tensor; then performing dynamic topology quantification processing on the data, establishing a multi-mode coupling network to evaluate the cracking state, and generating targeted repair decision parameters; then constructing a structural damage constraint function, and calculating a dynamic equilibrium boundary of the cracking area; and finally, based on the damage feature tensor and balance boundary optimization repair parameters, generating a targeted repair execution sequence. In addition, the repairing process is dynamically adjusted through closed-loop control. The method can accurately detect and perform targeted diagnosis and treatment on the cracking problem of the underground structure in the hot and humid areas, improves the pertinence and effectiveness of repair, guarantees the safe and stable operation of the underground structure, and has a remarkable practical value.

Description

Technical Field

[0001] The present invention relates to the technical field of underground structure engineering repair, and specifically to a targeted diagnosis and treatment method for underground structure cracking in hot and humid regions combined with non-destructive testing. Background Art

[0002] In modern urban construction, underground structures are widely used in fields such as subways, underground parking lots, underground shopping malls, and various municipal infrastructure. Their safety and stability are directly related to the normal operation of the city and the life and property safety of the public. Due to its special climate environment, underground structures in hot and humid regions face more severe challenges.

[0003] In hot and humid regions, it is hot and humid all year round, and this environment accelerates the aging process of underground structure materials. Taking concrete structures as an example, under the long-term action of heat and humidity, the cement hydration products in concrete will undergo a carbonation reaction with carbon dioxide in the air, reducing the alkalinity of the concrete, damaging the passivation film on the surface of the steel bars, and causing the steel bars to rust. After the steel bars rust, their volume expands, which will exert extrusion pressure on the surrounding concrete, thereby causing concrete cracking. At the same time, the hot and humid environment also promotes the growth and reproduction of microorganisms, and the metabolic products of microorganisms may erode the concrete, further weakening its structural performance.

[0004] The groundwater level is usually high in hot and humid regions, and underground structures are long-term soaked and permeated by groundwater. Groundwater contains various chemical components, such as sulfates, chlorides, etc. These substances will react with the components in the concrete to produce expansive products, causing cracking and damage to the concrete structure. For example, sulfates react with calcium hydroxide in the concrete to form gypsum, and the gypsum then reacts with tricalcium aluminate in the cement to form ettringite. The volume expansion of ettringite can cause cracks to occur inside the concrete structure. Moreover, the seepage of groundwater may also cause uneven settlement of underground structures, increasing the risk of structural cracking.

[0005] Traditional methods for detecting and repairing underground structure cracking have many limitations in hot and humid regions. In terms of detection, the commonly used visual inspection method is difficult to detect internal early damage and fine cracks, while the partial damage detection method can obtain relatively accurate internal information, but it will damage the structure, affect its normal use, and has low detection efficiency, unable to meet the needs of large-area detection. In terms of repair, traditional repair methods often lack pertinence, mostly adopt a unified repair plan, and do not fully consider the causes of cracks in different parts, the degree of development, and the actual stress conditions of the structure. This may not only lead to poor repair effects and be unable to effectively prevent the recurrence of cracks, but also cause waste of resources.

[0006] Due to the lack of effective detection and repair measures, once the problem of underground structure cracking occurs in humid and hot regions, if not dealt with in time, it will gradually deteriorate over time. In severe cases, it may even lead to the collapse of the structure. This will not only affect the normal use of the underground space, causing huge economic losses, but also may trigger safety accidents and threaten people's lives. Therefore, it is urgent to develop a method that can accurately detect and target the diagnosis and treatment of underground structure cracking problems in humid and hot regions. Summary of the Invention

[0007] The purpose of the present invention is to provide a method for targeted diagnosis and treatment of underground structure cracking in humid and hot regions combined with non-destructive testing to solve the problems raised in the above background technology.

[0008] To achieve the above purpose, the present invention provides the following technical solution: A method for targeted diagnosis and treatment of underground structure cracking in humid and hot regions combined with non-destructive testing, the method includes: Collect three-dimensional deformation data, crack distribution parameters, environmental temperature and humidity signals, and material stress response signals of the underground structure, and construct a structural damage feature tensor; Perform dynamic topological quantization processing on the structural damage feature tensor, generate a damage evolution parameter matrix, and establish a multi-modal coupling network for cracking state assessment to generate targeted repair decision parameters; Construct a structural damage constraint function according to the targeted repair decision parameters, and calculate the dynamic balance boundary of the cracking area; Based on the structural damage feature tensor and the dynamic balance boundary, perform iterative optimization of repair parameters to generate a targeted repair execution sequence.

[0009] Preferably, the collecting three-dimensional deformation data, crack distribution parameters, environmental temperature and humidity signals, and material stress response signals of the underground structure, and constructing a structural damage feature tensor includes: Perform multi-source synchronous collection on three-dimensional deformation data, crack distribution parameters, environmental temperature and humidity signals, and material stress response signals to generate a non-destructive testing raw data matrix; Perform spatial domain normalization processing on the raw data matrix to obtain a standardized data matrix, and perform modal separation on the standardized data matrix based on the tensor decomposition algorithm to generate initial damage feature components; Perform time domain alignment correction on the initial damage feature components, and perform phase compensation by introducing an environmental coupling coefficient to generate corrected damage feature components; Based on the multi-scale fusion algorithm, perform correlation analysis on the corrected damage feature components and the material degradation matrix to generate a damage correlation tensor, and perform multi-dimensional feature extraction on the damage correlation tensor to generate a structural damage feature tensor under different working conditions.

[0010] Preferably, the dynamic topological quantization process is performed on the structural damage feature tensor to generate a damage evolution parameter matrix, and a multi-modal coupling network is established for cracking state assessment to generate targeted repair decision parameters, including: Perform regional block quantization on the structural damage feature tensor to generate damage distribution interval parameters, and perform dynamic threshold calibration on the structural damage features based on the damage distribution interval parameters to generate a damage calibration matrix; Perform multi-level division processing on the damage feature data in the structural damage feature tensor according to the damage calibration matrix to generate a block feature matrix; Perform weighted topology processing on the block feature matrix and set a dynamic evolution interval to generate a topology feature matrix; Suppress environmental noise of the topology feature matrix through a non-linear filter to generate a noise reduction feature matrix, and perform tensor normalization conversion on the noise reduction feature matrix to generate a damage evolution parameter matrix; Based on the damage evolution parameter matrix, construct a multi-modal coupling network, and perform real-time assessment on the cracking damage state of the underground structure to generate targeted repair decision parameters.

[0011] Preferably, based on the damage evolution parameter matrix, construct a multi-modal coupling network, and perform real-time assessment on the cracking damage state of the underground structure to generate targeted repair decision parameters, including: Divide the damage evolution parameter matrix into a repair feature matrix and an assessment feature matrix; Construct a repair decision network based on the repair feature matrix. The repair decision network includes an input layer, a hidden layer, and an output layer. The input layer contains 4 neurons corresponding to deformation, crack, temperature and humidity, and stress signals. The hidden layer uses a piecewise non-linear activation function, and the output layer generates a repair decision value to obtain repair network output data; Construct a damage assessment network based on the assessment feature matrix. The damage assessment network includes an input layer, a hidden layer, and an output layer. The input layer receives the repair network output data. The hidden layer uses a hyperbolic tangent activation function, and the output layer generates a damage quantization value to obtain assessment network output data; Jointly train the repair decision network and the damage assessment network to generate a multi-modal coupling network structure, and perform residual function calculation and network robustness analysis on the multi-modal coupling network structure to generate a network assessment result; Based on the network assessment result, dynamically update the parameters of the repair decision network and the damage assessment network in the multi-modal coupling network structure, and adjust the network weights through the backpropagation algorithm to generate targeted repair decision parameters.

[0012] Preferably, constructing a structural damage constraint function based on the target repair decision parameters and calculating the dynamic equilibrium boundary of the cracking area includes: Construct an asymmetric matrix for the target repair decision parameters and generate a damage state matrix through linear transformation; Design a dynamic integral constraint term based on the damage state matrix, perform piecewise weight operation on the damage variables, generate a balance constraint term, and combine the damage state matrix and the balance constraint term to construct a structural stability function to generate a structural damage constraint function; Calculate the spatial gradient of the structural damage constraint function to generate a gradient analysis result, and perform a comparison operation between the gradient analysis result and the norm of the damage vector to generate a damage constraint condition; Divide the repair strength interval of the cracking area according to the damage constraint condition, set a dynamic balance coefficient to generate a constraint range, and perform boundary fitting based on the constraint range to generate the dynamic equilibrium boundary of the cracking area.

[0013] Preferably, iteratively optimizing the repair parameters based on the structural damage feature tensor and the dynamic equilibrium boundary to generate a target repair execution sequence includes: Construct a repair weight matrix and a material compensation matrix according to the structural damage feature tensor, and substitute the repair weight matrix and the material compensation matrix into a dynamic optimization function to generate a repair objective function; Apply material strength constraint conditions and repair rate limit conditions to the repair objective function, and convert the dynamic equilibrium boundary into a gradient convergence constraint to generate optimization constraint conditions; Input the repair objective function and the optimization constraint conditions into a numerical iterative solver to calculate the repair parameters in the discrete space to generate an initial sequence of repair parameters; Adjust the adaptive step size of the initial sequence of repair parameters, and perform stress constraint processing in combination with the material properties of the underground structure to generate a target repair execution sequence.

[0014] Preferably, the method further includes: Convert the target repair execution sequence into a construction control instruction, and perform error compensation on the construction control instruction to generate a repair execution signal; Calculate the target repair strength according to the stability requirements of the underground structure, and monitor the crack propagation rate in real time to generate crack monitoring data; Calculate the difference between the crack monitoring data and the target repair strength to generate a repair deviation value, and compare the repair deviation value with a preset repair threshold to generate a deviation determination result; Dynamically correct the repair parameters based on the deviation determination result, generate updated repair parameters, perform construction sequence conversion on the updated repair parameters, convert the repair parameters into construction strength values through material property mapping, and generate a new repair execution sequence; Input the new repair execution sequence into the repair execution device for closed-loop control, dynamically adjust the construction process, and complete the targeted repair of the underground structure cracking.

[0015] Preferably, the weighted topological processing of the block feature matrix and setting the dynamic evolution interval to generate the topological feature matrix includes: Perform regional weight allocation on the block feature matrix according to the material aging coefficient to generate a weighted feature matrix; Dynamically correct the threshold of the weighted feature matrix based on the environmental temperature and humidity data to generate a corrected feature matrix; Perform time-domain integration on the corrected feature matrix through the evolution rate function to generate a damage accumulation matrix; Compare the damage accumulation matrix with the preset critical threshold point by point to generate a dynamic evolution interval identification matrix.

[0016] Preferably, inputting the repair objective function and the optimization constraint conditions into the numerical iterative solver to calculate the repair parameters in the discrete space and generate the initial repair parameter sequence includes: Discretize the repair objective function using the finite difference method to generate a discrete optimization equation; Construct the Jacobian matrix of the constraint conditions based on the dynamic balance boundary to generate constraint gradient information; Numerically solve the discrete optimization equation through the Newton iteration algorithm to generate the initial value of the repair parameters; Conduct boundary compliance inspection on the initial value of the repair parameters, eliminate the parameters that exceed the constraint range, and generate the initial repair parameter sequence.

[0017] Preferably, converting the repair parameters into construction strength values through material property mapping to generate a new repair execution sequence includes: Establish a parameter-strength conversion function according to the material elastic modulus to generate a strength conversion coefficient matrix; Perform element-by-element multiplication operation on the repair parameters and the strength conversion coefficient matrix to generate a preliminary strength value sequence; Limit the amplitude of the preliminary strength value based on the maximum load-bearing capacity of the construction equipment to generate a standardized strength sequence; Perform time-series smoothing filtering on the standardized strength sequence to generate a new repair execution sequence.

[0018] Compared with the prior art, the beneficial effects of the present invention are: In terms of detection, by synchronously collecting multi-source data including three-dimensional deformation data, crack distribution parameters, ambient temperature and humidity signals, and material stress response signals, the state information of the underground structure can be comprehensively obtained. This multi-source data collection method avoids the limitations of single detection methods, making the detection results more accurate and reliable. After a series of operations such as spatial domain normalization, modal separation, and time domain alignment correction on the collected data, the generated structural damage feature tensor can accurately reflect the damage characteristics of the underground structure, and even minor damage changes can be effectively captured, providing a solid data basis for subsequent crack state assessment.

[0019] In terms of crack state assessment and repair decision-making, by performing dynamic topological quantization on the structural damage feature tensor, the generated damage evolution parameter matrix can clearly show the development trend of structural damage. Based on this, a multi-modal coupling network is constructed. By jointly training the repair decision network and the damage assessment network, real-time and accurate assessment of the crack damage state of the underground structure is achieved, and targeted repair decision parameters are generated. This process fully considers the mutual relationships among various factors, making the assessment results more in line with the actual situation, the repair decision more targeted, avoiding the problem of blind repair in traditional methods, and greatly improving the repair efficiency and quality.

[0020] During the repair implementation process, a structural damage constraint function is constructed according to the targeted repair decision parameters, and the dynamic balance boundary of the cracked area is calculated. The determination of this boundary provides a clear scope and standard for the repair work, ensuring that the repair process is carried out under reasonable mechanical conditions, and avoiding the situations of over-repair or under-repair. By iteratively optimizing the repair parameters, a targeted repair execution sequence is generated and adjusted in combination with material properties and construction actual situations, enabling the repair work to better meet the specific needs of the underground structure and further improving the effectiveness and reliability of the repair.

[0021] This method also sets up a closed-loop control link. By comparing and analyzing the crack monitoring data with the target repair strength, repair deviations are timely detected and the repair parameters are dynamically corrected, realizing real-time monitoring and adjustment of the repair process. This closed-loop control mechanism can ensure that the repair work always proceeds towards the expected goal, effectively improving the repair success rate and guaranteeing the long-term stability and safety of the underground structure. Generally speaking, the present invention can significantly improve the accuracy and effectiveness of diagnosing and treating cracks in underground structures in humid and hot regions, reduce maintenance costs, ensure the safe operation of underground structures, and has important engineering application value and social benefits. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 It is the working principle diagram of the method for targeted diagnosis and treatment of cracks in underground structures in humid and hot regions combined with non-destructive testing according to the present invention; Figure 2Flow chart for generating damage evolution parameter matrix for dynamic topology quantization processing; Figure 3 Flow chart for generating targeted repair decision parameters for establishing a multi-modal coupling network to evaluate the cracking state; Figure 4 Flow chart for constructing a structural damage constraint function to calculate the dynamic balance boundary of the cracking area; Figure 5 Flow chart for process conversion and subsequent closed-loop control of the targeted repair execution sequence. Specific implementation manners

[0023] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0024] Please refer to Figures 1 - 5 , the present invention provides a method for targeted diagnosis and treatment of underground structure cracking in humid and hot regions combined with non-destructive testing, and the specific implementation steps are as follows: Use professional non-destructive testing equipment to synchronously collect three-dimensional deformation data, crack distribution parameters, environmental temperature and humidity signals, and material stress response signals of the underground structure. These data reflect the state of the underground structure from different angles and are the basis for subsequent analysis. The collected data forms a non-destructive testing raw data matrix, and then it is subjected to spatial domain normalization processing to eliminate the differences caused by different dimensions and value ranges between the data for subsequent analysis. Modal separation is performed on the standardized data through a tensor decomposition algorithm to obtain the initial components of damage characteristics. Due to possible slight differences in data collection time, time domain alignment correction is performed on the initial components of damage characteristics, and an environmental coupling coefficient is introduced for phase compensation to generate more accurate damage characteristic components. Finally, based on a multi-scale fusion algorithm, the corrected damage characteristic components are correlated with the material degradation matrix to extract multi-dimensional features and generate a structural damage characteristic tensor under different working conditions to comprehensively characterize the damage characteristics of the underground structure.

[0025] Regionally block and quantify the structural damage feature tensor, determine the damage distribution interval parameters, and accordingly perform dynamic threshold calibration on the structural damage features to generate a damage calibration matrix. According to the damage calibration matrix, hierarchically divide the data in the structural damage feature tensor to obtain a block feature matrix. Allocate regional weights to the block feature matrix according to the material aging coefficient, and then dynamically correct the threshold in combination with the environmental temperature and humidity data. Through operations such as time-domain integration of the evolution rate function and comparison with the preset critical threshold, generate a topological feature matrix. Suppress environmental noise and perform tensor normalization transformation on the topological feature matrix to obtain a damage evolution parameter matrix. Based on the damage evolution parameter matrix, construct a multi-modal coupling network. Divide the matrix into a repair feature matrix and an evaluation feature matrix, and respectively construct a repair decision network and a damage evaluation network. The input layer of the repair decision network corresponds to deformation, crack, temperature and humidity, and stress signals. The hidden layer uses a piecewise non-linear activation function, and the output layer generates a repair decision value. The input layer of the damage evaluation network receives the output data of the repair network. The hidden layer uses a hyperbolic tangent activation function, and the output layer generates a damage quantification value. Jointly train the two networks, and after calculating the residual function and analyzing the network robustness, dynamically update the network parameters, and finally generate targeted repair decision parameters.

[0026] Construct an asymmetric matrix according to the targeted repair decision parameters, and obtain a damage state matrix through linear transformation. Design a dynamic integral constraint term based on the damage state matrix, perform piecewise weighted operations on the damage variables to generate a balance constraint term, combine the two to construct a structural stability function, and then obtain a structural damage constraint function. Calculate the spatial gradient of the structural damage constraint function, compare it with the norm of the damage vector, and generate a damage constraint condition. Divide the repair strength interval of the cracking area according to the damage constraint condition, set a dynamic balance coefficient to determine the constraint range, and obtain the dynamic balance boundary of the cracking area through boundary fitting, providing an important boundary basis for repair.

[0027] Construct a repair weight matrix and a material compensation matrix based on the structural damage feature tensor, and substitute them into the dynamic optimization function to generate a repair objective function. Apply material strength and repair rate limit conditions to the repair objective function, and transform the dynamic balance boundary into a gradient convergence constraint to form an optimization constraint condition. Discretize the repair objective function using the finite difference method, construct a Jacobian matrix of the constraint conditions to obtain gradient information, and solve through the Newton iteration algorithm to obtain the initial value of the repair parameters. After passing the boundary compliance test, obtain the initial sequence of the repair parameters. Perform adaptive step size adjustment on the initial sequence, and perform stress constraint processing in combination with the material properties of the underground structure, and finally generate a targeted repair execution sequence, providing specific operation steps for actual repair.

[0028] The technical solution of the present invention will be further described in detail below through specific embodiments.

[0029] Example 1

[0030] In this embodiment, a detailed description is given of the process of collecting three-dimensional deformation data, crack distribution parameters, environmental temperature and humidity signals, and material stress response signals of the underground structure, and constructing the structural damage feature tensor.

[0031] Professional non-destructive testing equipment is used to collect three-dimensional deformation data, crack distribution parameters, environmental temperature and humidity signals, and material stress response signals simultaneously. These devices can be high-precision laser measuring instruments for obtaining three-dimensional deformation data, crack observation instruments for measuring crack distribution parameters, temperature and humidity sensors for collecting environmental temperature and humidity signals, and stress sensors for collecting material stress response signals. Through multi-source synchronous acquisition, an original non-destructive testing data matrix is generated.

[0032] The original data matrix is subjected to spatial domain normalization. The purpose of this step is to eliminate the differences in dimension and numerical range among different data sources, making the data comparable. After processing, a standardized data matrix is obtained, and then based on the tensor decomposition algorithm, modal separation is performed on the standardized data matrix, and the complex data is decomposed into multiple initial damage feature components.

[0033] The initial damage feature components are corrected for time domain alignment. Since there may be slight differences in the acquisition time of different types of data, by introducing an environmental coupling coefficient for phase compensation, it is ensured that each component can accurately correspond in the time dimension, generating corrected damage feature components.

[0034] Finally, based on the multi-scale fusion algorithm, a correlation analysis is performed between the corrected damage feature components and the material degradation matrix. The material degradation matrix reflects the performance degradation of the material over time and under different environmental conditions. Through the correlation analysis, a damage correlation tensor is generated, and then multi-dimensional feature extraction is performed on the damage correlation tensor, thereby generating the structural damage feature tensor under different working conditions, providing a comprehensive and accurate data basis for subsequent analysis and decision-making.

[0035] Suppose there is an underground parking lot located in a hot and humid area. Due to long-term environmental influence, signs of structural cracking have appeared. A professional three-dimensional laser scanner is used to scan the structure of the underground parking lot, and data is collected every certain period (such as 1 hour) to obtain the displacement changes of different positions in three-dimensional space, thereby obtaining three-dimensional deformation data. Using a high-precision crack observation instrument, carefully measure the length, width, spacing and other parameters of the cracks along the beams, columns and walls of the parking lot, and record the crack distribution parameters. Install multiple temperature and humidity sensors in different areas of the parking lot to collect environmental temperature and humidity signals in real time, for example, record the temperature and humidity values every 10 minutes. At the same time, stress sensors are arranged at key stress-bearing parts to obtain material stress response signals and monitor the internal stress changes of the structure. After these data are synchronously collected, an original non-destructive testing data matrix is formed.

[0036] In the original data matrix, the dimensions and numerical ranges of different types of data vary greatly. For example, the unit of three-dimensional deformation data is millimeters, the unit of crack width data is millimeters, the units of temperature and humidity data are degrees Celsius and percentage respectively, and the unit of stress data is megapascals. The original data is processed by spatial domain normalization to map various types of data to a unified interval (such as 0 - 1), obtaining a standardized data matrix. Using the tensor decomposition algorithm, the standardized data matrix is decomposed into multiple initial components of damage characteristics, just like "disassembling" complex structural damage data into more easily analyzable parts.

[0037] Because there may be slight differences in the data acquisition times of different types of data. For example, the data acquisition time of the stress sensor is a few seconds later than that of the temperature and humidity sensor. Therefore, it is necessary to perform time domain alignment correction on the initial components of damage characteristics. An environmental coupling coefficient is introduced, which comprehensively considers the influence of temperature, humidity, material properties, etc. on the data acquisition time difference, and performs phase compensation on each component to make them accurately correspond in time, generating corrected damage characteristic components.

[0038] The material degradation matrix is constructed based on the research data of the performance attenuation of parking lot building materials (such as concrete, steel) in a humid and hot environment for many years. Based on the multi-scale fusion algorithm, the corrected damage characteristic components are associated with the material degradation matrix for analysis, generating a damage correlation tensor. Multidimensional features are extracted from this tensor, such as the difference in damage degrees in different regions in the spatial dimension and the damage development trend in the time dimension. Finally, a structural damage characteristic tensor applicable to different working conditions (such as normal use, after heavy rain, etc.) of this underground parking lot is generated.

[0039] Example 2

[0040] This example details the process of performing dynamic topological quantization processing on the structural damage characteristic tensor, generating a damage evolution parameter matrix, and establishing a multi-modal coupling network for crack state assessment to generate targeted repair decision parameters.

[0041] First, perform regional block quantization on the structural damage characteristic tensor. Divide the entire structural damage characteristic tensor into multiple regions according to certain rules, and calculate the damage distribution interval parameters of each region. Based on these parameters, perform dynamic threshold calibration on the structural damage characteristics to determine the threshold range corresponding to different damage degrees, generating a damage calibration matrix.

[0042] Subsequently, the damage characteristic data in the structural damage characteristic tensor is processed by multi-level division according to the damage calibration matrix. According to different levels of damage degrees, the data is classified to generate a block feature matrix.

[0043] Next, regional weight distribution is performed on the block feature matrix according to the material aging coefficient. The material aging coefficient reflects the attenuation of material properties over time, and higher weights are assigned to regions with a high degree of aging. After generating the weighted feature matrix, dynamic threshold correction is performed based on the environmental temperature and humidity data. The changes in temperature and humidity will affect the development of structural damage, and the threshold is adjusted according to the real-time temperature and humidity data to generate the corrected feature matrix.

[0044] Then, the corrected feature matrix is integrated in the time domain through the evolution rate function to calculate the cumulative situation of damage in the time dimension, generating the damage accumulation matrix. The damage accumulation matrix is compared point by point with the preset critical threshold to determine whether each point reaches or exceeds the threshold, generating the dynamic evolution interval identification matrix, and then generating the topological feature matrix.

[0045] The topological feature matrix is subjected to environmental noise suppression through a nonlinear filter to remove noise interference in the data, generating the noise-reduced feature matrix. Tensor normalization transformation is performed on the noise-reduced feature matrix to meet the requirements of subsequent analysis, generating the damage evolution parameter matrix.

[0046] Based on the damage evolution parameter matrix, a multi-modal coupling network is constructed. The damage evolution parameter matrix is divided into a repair feature matrix and an evaluation feature matrix, and a repair decision network and a damage evaluation network are constructed respectively. The input layer of the repair decision network contains 4 neurons corresponding to deformation, crack, temperature and humidity, and stress signals. The hidden layer uses a piecewise non-linear activation function, and the output layer generates a repair decision value to obtain the output data of the repair network. The input layer of the damage evaluation network receives the output data of the repair network. The hidden layer uses a hyperbolic tangent activation function, and the output layer generates a damage quantification value to obtain the output data of the evaluation network.

[0047] The repair decision network and the damage evaluation network are jointly trained to optimize the network parameters, generating the multi-modal coupling network structure. Residual function calculation and network robustness analysis are performed on the multi-modal coupling network structure to evaluate the network performance. Based on the evaluation results, the parameters of the repair decision network and the damage evaluation network in the network are dynamically updated, and the network weights are adjusted through the backpropagation algorithm to finally generate the targeted repair decision parameters.

[0048] Taking a subway tunnel located in a hot and humid area as an example, different degrees of structural damage have occurred during the long-term operation of the tunnel. First, the structural damage feature tensor of the subway tunnel is divided into multiple regions according to a certain length (such as 10 meters), and the damage distribution interval parameters in each region are calculated, such as the maximum value, minimum value and occurrence frequency of the crack width in a certain region. According to these parameters, the thresholds corresponding to different damage degrees are determined. For example, a crack width greater than 1 mm is severe damage, 0.5 - 1 mm is moderate damage, and less than 0.5 mm is mild damage, so as to generate the damage calibration matrix.

[0049] According to the damage calibration matrix, the data in the tunnel structure damage feature tensor are hierarchically divided. For example, data such as crack width and deformation amount are classified according to the standards of mild, moderate, and severe damage to form data blocks at different levels, and then a block feature matrix is generated.

[0050] Determine the material aging coefficient based on the service life of the subway tunnel and the research data on material aging. For areas with a longer service life, the material aging coefficient is higher. Perform regional weight allocation on the block feature matrix, and assign a higher weight to the area with a higher aging coefficient. Then, dynamically correct the weighted feature matrix by combining the real-time temperature and humidity data in the tunnel. If the humidity in the tunnel suddenly increases, appropriately reduce the damage threshold of the crack width because the increase in humidity may accelerate the development of cracks, and generate a corrected feature matrix.

[0051] Perform an integral operation on the corrected feature matrix in the time dimension through the set evolution rate function. For example, calculate the daily damage accumulation amount at one-day intervals to generate a damage accumulation amount matrix. Compare this matrix point by point with a preset critical threshold (such as when the damage accumulation amount in a certain area exceeds a certain value, it needs to be focused on) to determine whether each area reaches or exceeds the threshold, generate a dynamic evolution interval identification matrix, and then obtain a topological feature matrix.

[0052] Process the topological feature matrix using a non-linear filter to remove environmental noise caused by sensor errors, etc., and generate a noise-reduced feature matrix. Then, perform tensor normalization conversion on the noise-reduced feature matrix to make the data meet the requirements of subsequent analysis, and obtain a damage evolution parameter matrix.

[0053] Divide the damage evolution parameter matrix into a repair feature matrix and an evaluation feature matrix. Build a repair decision network based on the repair feature matrix. The 4 neurons in the input layer correspond to the deformation, crack, temperature and humidity, and stress signals of the tunnel respectively. The hidden layer uses a piecewise non-linear activation function, such as the ReLU function, to enable the network to better learn the data features. The output layer generates a repair decision value, such as the recommended repair method (whether it is local repair or overall reinforcement, etc.), and obtains the output data of the repair network. Build a damage evaluation network based on the evaluation feature matrix. The input layer receives the output data of the repair network. The hidden layer uses the hyperbolic tangent activation function, and the output layer generates a damage quantification value to reflect the severity of the current tunnel damage, and obtains the output data of the evaluation network.

[0054] Jointly train the repair decision network and the damage assessment network, continuously adjust the network parameters, and generate a multi-modal coupled network structure. Evaluate the difference between the network prediction value and the actual value by calculating the residual function, and conduct network robustness analysis. According to the analysis results, use the backpropagation algorithm to dynamically update the network weights, and finally generate targeted repair decision parameters for the subway tunnel, such as determining the key repair areas and the amount of repair materials used.

[0055] Example 3

[0056] This embodiment focuses on the implementation process of constructing a structural damage constraint function according to the targeted repair decision parameters and calculating the dynamic balance boundary of the cracking area.

[0057] Construct an asymmetric matrix for the targeted repair decision parameters. According to the characteristics and mutual relationships of the decision parameters, generate a damage state matrix through a specific linear transformation, which can intuitively reflect the current damage state of the underground structure.

[0058] Design a dynamic integral constraint term based on the damage state matrix. Perform piecewise weight operations on the damage variables, assign different weights according to different degrees of damage, and generate a balance constraint term. Combine the damage state matrix and the balance constraint term to construct a structural stability function, thereby generating a structural damage constraint function, which comprehensively considers the damage situation and stability requirements of the structure.

[0059] Calculate the spatial gradient of the structural damage feature tensor to obtain the gradient analysis result. Compare the gradient analysis result with the norm of the damage vector to determine the spatial variation of the damage and generate damage constraint conditions.

[0060] Divide the repair strength interval of the cracking area according to the damage constraint conditions. Set a dynamic balance coefficient according to the actual situation of the structure to determine the constraint range. Based on the constraint range, perform boundary fitting and use a suitable mathematical method, such as a curve fitting algorithm, to generate the dynamic balance boundary of the cracking area. This boundary clarifies the strength range and boundary conditions required for repair under different damage conditions, providing an important basis for subsequent repair work.

[0061] Suppose there is an underground warehouse built for many years in a hot and humid area, and its structure has cracked problems. It is necessary to determine the dynamic balance boundary of the cracking area. The obtained targeted repair decision parameters, such as the strength requirements of the repair materials and the repair time requirements, are used to construct an asymmetric matrix through a specific linear transformation to form a damage state matrix. It is expressed by the formula: , where represents the damage state matrix, are the targeted repair decision parameters, Represents a specific linear transformation operation, which is determined according to the relationship between the characteristics of the underground warehouse structure and the repair decision parameters.

[0062] Based on the damage state matrix , perform piecewise weight operations on damage variables (such as crack width, structural deformation, etc.). For example, assign a lower weight to areas where the crack width is less than 0.5 mm and a higher weight to areas greater than 0.5 mm to generate a balance constraint term. Combine the damage state matrix and the balance constraint term to construct a structural stability function, that is, a structural damage constraint function .

[0063] Calculate the spatial gradient of the structural damage constraint function to obtain the gradient analysis result. Compare this result with the magnitude of the damage vector (a vector composed of crack width, deformation, etc.) to determine the spatial variation of the damage and obtain the damage constraint condition.

[0064] According to the damage constraint condition, divide the repair strength interval of the cracked area of the underground warehouse. For example, it is divided into a low-strength repair area, a medium-strength repair area, and a high-strength repair area. Set a dynamic balance coefficient, which comprehensively considers factors such as the usage requirements of the warehouse and the structural bearing capacity. Determine the constraint range according to the partition and the dynamic balance coefficient, and use the curve fitting algorithm to fit the boundary of the constraint range, and finally generate the dynamic balance boundary of the cracked area, providing a key basis for subsequent repairs.

[0065] Example 4

[0066] This example focuses on describing the specific operations of iteratively optimizing repair parameters based on the structural damage feature tensor and the dynamic balance boundary to generate a targeted repair execution sequence.

[0067] Construct a repair weight matrix and a material compensation matrix according to the structural damage feature tensor. The repair weight matrix reflects the importance of different damaged parts to the repair work, and the material compensation matrix takes into account the differences in material properties. Substitute the repair weight matrix and the material compensation matrix into the dynamic optimization function to generate a repair objective function, which aims to achieve the optimal repair of the underground structure.

[0068] Apply material strength constraint conditions and repair rate limit conditions to the repair objective function. The material strength constraint conditions ensure that the material strength used in the repair process meets the structural requirements, and the repair rate limit conditions avoid the repair work being too hasty or slow, affecting the repair effect and the project progress. At the same time, convert the dynamic balance boundary into a gradient convergence constraint to generate optimization constraint conditions, further restricting the value range of the repair parameters.

[0069] Input the repair objective function and optimization constraints into a numerical iterative solver. Use the finite difference method to discretize the repair objective function, transforming the continuous function into a discrete equation form to generate a discrete optimization equation. Construct the Jacobian matrix of the constraints based on the dynamic balance boundary to obtain the constraint gradient information. Numerically solve the discrete optimization equation through the Newton iteration algorithm to obtain the initial values of the repair parameters. Conduct a boundary compliance check on the initial values of the repair parameters, eliminating the parameters that exceed the constraint range to generate the initial sequence of repair parameters.

[0070] Adjust the initial sequence of repair parameters with an adaptive step size, dynamically adjusting the iteration step size according to the actual situation to improve the calculation efficiency and accuracy. Conduct stress constraint processing in combination with the material properties of the underground structure to ensure that the stress state of the structure during the repair process is within the safe range, and finally generate a targeted repair execution sequence to provide specific operation steps and parameters for the actual repair work.

[0071] Take an underground sewage treatment tank located in a hot and humid area as an example. Cracks have appeared in the treatment tank and need to be repaired. Construct a repair weight matrix and a material compensation matrix based on the structural damage feature tensor of the sewage treatment tank. For example, in the area near the sewage inlet and outlet, due to the scouring and corrosion of the water flow, the damage is relatively serious, and a higher weight is assigned in the repair weight matrix; for the special anti-corrosion materials used in the tank body, construct a material compensation matrix according to their performance. Substitute these two matrices into the dynamic optimization function to generate a repair objective function, with the goal of making the sewage treatment tank reach the best performance state after repair.

[0072] Impose material strength constraint conditions and repair rate limit conditions on the repair objective function. The material strength constraint condition requires that the strength of the repair material used is not lower than the strength of the original material of the tank body. The repair rate limit condition stipulates that the area repaired per day cannot exceed a certain value to avoid affecting the quality due to too fast repair or delaying the construction period due to too slow repair. At the same time, convert the dynamic balance boundary into a gradient convergence constraint to form optimization constraint conditions.

[0073] Input the repair objective function and optimization constraints into a numerical iterative solver. Use the finite difference method to discretize the repair objective function, turning the continuous function into discrete points for calculation to obtain a discrete optimization equation. Construct the Jacobian matrix of the constraints based on the dynamic balance boundary to obtain the constraint gradient information. Numerically solve the discrete optimization equation using the Newton iteration algorithm to obtain the initial values of the repair parameters. Conduct a boundary compliance check on these initial values, eliminating the parameters that exceed the constraint range, such as parameters that exceed the material strength range or the repair rate range, to generate the initial sequence of repair parameters.

[0074] Adaptive step - size adjustment is performed on the initial sequence of repair parameters. In the initial stage of iteration, the step - size can be set relatively large to accelerate the calculation speed; as it approaches the optimal solution, the step - size is reduced to improve the calculation accuracy. Stress constraints are processed in combination with the material properties of the sewage treatment tank to ensure that the stress of the tank body is within the safe range during the repair process. Finally, a targeted repair execution sequence is generated, specifying the specific steps and parameters of the repair, such as the amount of repair material used in each area, the repair order, etc.

[0075] Example 5

[0076] This example details the subsequent process of completing the targeted repair of underground structure cracking after generating the targeted repair execution sequence.

[0077] Process conversion is carried out on the targeted repair execution sequence to convert the repair parameters into specific construction operation instructions, such as construction sequence, construction method, etc. Error compensation is performed on the construction control instructions. Considering various error factors that may occur during the actual construction process, compensation is carried out through a certain algorithm to generate a repair execution signal.

[0078] Calculate the target repair strength quantity according to the stability requirements of the underground structure. This strength quantity is a key indicator to ensure that the structure can operate stably after repair. At the same time, the crack propagation rate is monitored in real - time, and real - time data on crack propagation is obtained through professional monitoring equipment to generate crack monitoring data.

[0079] Calculate the difference between the crack monitoring data and the target repair strength quantity to obtain a repair deviation value. Compare the repair deviation value with a preset repair threshold to generate a deviation determination result. If the deviation value exceeds the threshold, it indicates that there are problems with the repair work and adjustments are needed.

[0080] Dynamically correct the repair parameters based on the deviation determination result. If the deviation exceeds the range, adjust the repair parameters using a suitable algorithm according to the magnitude and direction of the deviation to generate updated repair parameters. Perform construction sequence conversion on the updated repair parameters, establish a parameter - strength conversion function based on the material elastic modulus to generate a strength conversion coefficient matrix. Perform element - by - element multiplication operation on the repair parameters and the strength conversion coefficient matrix to generate a preliminary strength value sequence. Limit the amplitude of the preliminary strength value based on the maximum load - bearing capacity of the construction equipment to ensure that the construction strength is within the range that the equipment can withstand, generating a standardized strength sequence. Perform time - series smoothing filtering on the standardized strength sequence to remove the fluctuations in the data and generate a new repair execution sequence.

[0081] Input the new repair execution sequence into the repair execution device for closed - loop control, and dynamically adjust the construction process according to the real - time monitored data. Continuously repeat the above process until the repair deviation value is within the preset threshold range, completing the targeted repair of underground structure cracking and ensuring the safety and stability of the underground structure.

[0082] Take an underground pipe gallery in a hot and humid area as an example. During routine inspections, it was found that some of the structures in the pipe gallery had cracks, and now it is being repaired. The generated targeted repair execution sequence is converted into a process. For example, if a section of the pipe gallery needs to be partially repaired in the repair execution sequence, it will be converted into specific construction operation instructions during the process conversion, including what kind of repair materials to use, what kind of repair process to adopt (such as grouting repair or smearing repair), and the construction personnel's operating procedures. Since there may be errors in the actual construction process, the construction control instructions are compensated for errors. For example, considering the precision errors of the construction equipment and the inconsistency of the construction personnel's operations, the instructions are adjusted through a specific algorithm to generate a repair execution signal.

[0083] The target repair strength is calculated based on the design bearing capacity and stability requirements of the underground pipe gallery. Assuming that the pipe gallery design requires that the structural deformation cannot exceed a certain value when subjected to a certain pressure, the strength value that the repaired structure should reach is calculated based on this requirement. At the same time, sensors are used to monitor the crack expansion rate in real time, and the crack expansion length is recorded every certain period of time (such as half an hour) to generate crack monitoring data.

[0084] Calculate the difference between the crack monitoring data and the target repair strength. For example, the crack width corresponding to the target repair strength should be stable within 0.2 mm, and the current crack width monitored is 0.3 mm. The difference between the two is the repair deviation value. Compare the repair deviation value with the preset repair threshold (such as 0.1 mm). If the deviation value exceeds the threshold, it means that there is a problem with the repair work and it needs to be adjusted.

[0085] The repair parameters are dynamically corrected based on the deviation judgment results. If the deviation exceeds the range, the repair parameters are adjusted using an appropriate algorithm based on the size and direction of the deviation. For example, if the crack width is too large, the amount of repair material should be appropriately increased. The updated repair parameters are converted into a construction sequence, and a parameter-strength conversion function is established based on the elastic modulus of the corridor building materials to generate a strength conversion coefficient matrix. The repair parameters are element-by-element multiplied with the strength conversion coefficient matrix to obtain a preliminary strength value sequence. Taking into account the maximum load-bearing capacity of the construction equipment, the preliminary strength value is limited. For example, the maximum strength that the construction equipment can apply is a certain value, and the excess is adjusted to generate a standardized strength sequence. The standardized strength sequence is time-series smoothed and filtered to remove data fluctuations and generate a new repair execution sequence.

[0086] Input the new repair execution sequence into the repair execution device, such as an automated patching device. Dynamically adjust the construction process according to the real-time monitored data. If it is found that the repair effect of a certain section of the pipe gallery is not good, timely adjust the injection volume of the repair material or the working frequency of the patching device. Continuously repeat the above process until the repair deviation value is within the preset threshold range, complete the targeted repair of the underground pipe gallery cracking, and ensure the safe and stable operation of the pipe gallery.

[0087] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or further includes elements inherent to such process, method, article or device.

[0088] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for targeted diagnosis and treatment of underground structure cracking in humid and hot regions combined with non-destructive testing, characterized in that, It includes the following steps: Collect three-dimensional deformation data, crack distribution parameters, environmental temperature and humidity signals, and material stress response signals of the underground structure, and construct a structural damage characteristic tensor; Perform dynamic topological quantization processing on the structural damage characteristic tensor, generate a damage evolution parameter matrix, and establish a multi-modal coupling network for cracking state assessment to generate targeted repair decision parameters; Construct a structural damage constraint function according to the targeted repair decision parameters, and calculate the dynamic balance boundary of the cracking area; Based on the structural damage characteristic tensor and the dynamic balance boundary, perform iterative optimization of repair parameters to generate a targeted repair execution sequence.

2. The method for targeted diagnosis and treatment of cracks in underground structures in humid and hot regions combined with non-destructive testing according to claim 1, wherein The collection of three-dimensional deformation data, crack distribution parameters, environmental temperature and humidity signals, and material stress response signals of the underground structure, and the construction of a structural damage characteristic tensor include: Perform multi-source synchronous acquisition on three-dimensional deformation data, crack distribution parameters, environmental temperature and humidity signals, and material stress response signals to generate a raw data matrix for non-destructive testing; Perform spatial domain normalization processing on the raw data matrix to obtain a standardized data matrix, and perform modal separation on the standardized data matrix based on a tensor decomposition algorithm to generate initial damage characteristic components; Perform time domain alignment correction on the initial damage characteristic components, and perform phase compensation by introducing an environmental coupling coefficient to generate corrected damage characteristic components; Based on a multi-scale fusion algorithm, perform correlation analysis on the corrected damage characteristic components and a material degradation matrix to generate a damage correlation tensor, and perform multi-dimensional feature extraction on the damage correlation tensor to generate a structural damage characteristic tensor under different working conditions.

3. The method for targeted diagnosis and treatment of cracks in underground structures in humid and hot regions combined with non-destructive testing according to claim 1, characterized in that, The dynamic topological quantization processing of the structural damage characteristic tensor to generate a damage evolution parameter matrix, and the establishment of a multi-modal coupling network for cracking state assessment to generate targeted repair decision parameters include: Perform regional block quantization on the structural damage characteristic tensor to generate damage distribution interval parameters, and perform dynamic threshold calibration on the structural damage characteristics based on the damage distribution interval parameters to generate a damage calibration matrix; Perform multi-level division processing on the damage characteristic data in the structural damage characteristic tensor according to the damage calibration matrix to generate a block feature matrix; Perform weighted topology processing on the block feature matrix and set a dynamic evolution interval to generate a topology feature matrix; Suppress environmental noise of the topology feature matrix through a non-linear filter to generate a noise reduction feature matrix, and perform tensor standardization conversion on the noise reduction feature matrix to generate a damage evolution parameter matrix; Based on the damage evolution parameter matrix, construct a multi-modal coupling network, and perform real-time assessment on the cracking damage state of the underground structure to generate targeted repair decision parameters.

4. The method for targeted diagnosis and treatment of cracks in underground structures in humid and hot regions combined with non-destructive testing according to claim 3, wherein, The construction of a multi-modal coupling network based on the damage evolution parameter matrix, and the real-time assessment of the cracking damage state of the underground structure to generate targeted repair decision parameters include: Divide the damage evolution parameter matrix into a repair feature matrix and an assessment feature matrix; Construct a repair decision network based on the repair feature matrix. The repair decision network includes an input layer, a hidden layer, and an output layer. The input layer contains 4 neurons corresponding to deformation, crack, temperature and humidity, and stress signals. The hidden layer uses a piecewise non-linear activation function, and the output layer generates a repair decision value to obtain the repair network output data; Construct a damage assessment network based on the evaluation feature matrix. The damage assessment network includes an input layer, a hidden layer, and an output layer. The input layer receives the repair network output data. The hidden layer uses a hyperbolic tangent activation function, and the output layer generates a damage quantification value to obtain the evaluation network output data; Jointly train the repair decision network and the damage assessment network to generate a multi-modal coupling network structure, and perform residual function calculation and network robustness analysis on the multi-modal coupling network structure to generate a network evaluation result; Dynamically update the parameters of the repair decision network and the damage assessment network in the multi-modal coupling network structure based on the network evaluation result, and adjust the network weights through the backpropagation algorithm to generate targeted repair decision parameters.

5. The method for targeted diagnosis and treatment of cracks in underground structures in humid and hot regions combined with non-destructive testing according to claim 1, characterized in that Construct a structural damage constraint function according to the targeted repair decision parameters and calculate the dynamic balance boundary of the cracked area, including: Construct an asymmetric matrix for the targeted repair decision parameters and generate a damage state matrix through linear transformation; Design a dynamic integral constraint term based on the damage state matrix, perform piecewise weight operation on the damage variable to generate a balance constraint term, and combine the damage state matrix and the balance constraint term to construct a structural stability function to generate a structural damage constraint function; Calculate the spatial gradient of the structural damage constraint function to generate a gradient analysis result, and perform a comparison operation between the gradient analysis result and the norm of the damage vector to generate a damage constraint condition; Partition the repair strength interval of the cracked area according to the damage constraint condition, set a dynamic balance coefficient to generate a constraint range, and perform boundary fitting based on the constraint range to generate the dynamic balance boundary of the cracked area.

6. The method for targeted diagnosis and treatment of cracks in underground structures in humid and hot regions combined with non-destructive testing according to claim 1, characterized in that, Iteratively optimize the repair parameters based on the structural damage feature tensor and the dynamic balance boundary to generate a targeted repair execution sequence, including: Construct a repair weight matrix and a material compensation matrix according to the structural damage feature tensor, and substitute the repair weight matrix and the material compensation matrix into the dynamic optimization function to generate a repair objective function; Apply material strength constraint conditions and repair rate limit conditions to the repair objective function, and convert the dynamic balance boundary into a gradient convergence constraint to generate optimization constraint conditions; Input the repair objective function and the optimization constraint conditions into a numerical iterative solver to calculate the repair parameters in the discrete space to generate an initial sequence of repair parameters; Adjust the adaptive step size of the initial sequence of repair parameters, and perform stress constraint processing in combination with the material properties of the underground structure to generate a targeted repair execution sequence.

7. The method for targeted diagnosis and treatment of cracks in underground structures in humid and hot regions combined with non-destructive testing according to claim 1, characterized in that, The method further includes: Convert the targeted repair execution sequence into a construction control instruction, and perform error compensation on the construction control instruction to generate a repair execution signal; Calculate the target repair intensity according to the stability requirements of the underground structure, and monitor the crack propagation rate in real time to generate crack monitoring data; Calculate the difference between the crack monitoring data and the target repair intensity to generate a repair deviation value, and compare the repair deviation value with a preset repair threshold to generate a deviation determination result; Dynamically correct the repair parameters based on the deviation determination result to generate updated repair parameters, and perform construction sequence conversion on the updated repair parameters. Convert the repair parameters into construction intensity values through material property mapping to generate a new repair execution sequence; Input the new repair execution sequence into the repair execution device for closed-loop control, dynamically adjust the construction process, and complete the targeted repair of the underground structure cracking.

8. The method for targeted diagnosis and treatment of cracks in underground structures in humid and hot regions combined with non-destructive testing according to claim 3, wherein The weighted topology processing of the block feature matrix and setting the dynamic evolution interval to generate the topology feature matrix includes: Perform regional weight assignment on the block feature matrix according to the material aging coefficient to generate a weighted feature matrix; Dynamically correct the weighted feature matrix based on the environmental temperature and humidity data to generate a corrected feature matrix; Perform time-domain integration on the corrected feature matrix through the evolution rate function to generate a damage accumulation matrix; Compare the damage accumulation matrix with a preset critical threshold point by point to generate a dynamic evolution interval identification matrix.

9. The method for targeted diagnosis and treatment of cracks in underground structures in humid and hot regions combined with non-destructive testing according to claim 6, wherein Input the repair objective function and the optimization constraint conditions into the numerical iterative solver to calculate the repair parameters in the discrete space and generate the initial sequence of repair parameters, including: Discretize the repair objective function using the finite difference method to generate a discrete optimization equation; Construct the Jacobian matrix of the constraint conditions based on the dynamic balance boundary to generate constraint gradient information; Numerically solve the discrete optimization equation through the Newton iteration algorithm to generate the initial value of the repair parameters; Conduct boundary compliance inspection on the initial value of the repair parameters, eliminate the parameters beyond the constraint range, and generate the initial sequence of repair parameters.

10. The method for targeted diagnosis and treatment of cracks in underground structures in humid and hot regions combined with non-destructive testing according to claim 7, wherein, Convert the repair parameters into construction intensity values through material property mapping to generate a new repair execution sequence, including: Establish a parameter-strength conversion function according to the material elastic modulus to generate a strength conversion coefficient matrix; Perform element-by-element multiplication operation on the repair parameters and the strength conversion coefficient matrix to generate a preliminary strength value sequence; Limit the preliminary strength value based on the maximum load-bearing capacity of the construction equipment to generate a standardized strength sequence; Perform time-series smoothing filtering on the standardized strength sequence to generate a new repair execution sequence.

Images