Bridge concrete disease layered detection and repair effect review method

Abstract

The application belongs to the technical field of bridge concrete disease detection and repair, and specifically discloses a bridge concrete disease layered detection and repair effect review method. The method fuses hyperspectral imaging and electromagnetic induction data through a double-channel mutual verification positioning mechanism to realize accurate layered identification of disease depth and physical properties. The method combines a steel reinforcement shielding layered correction model and a dynamic load interference filtering algorithm to improve detection reliability. A layered quantitative evaluation system including density uniformity, interface bonding strength gradient and residual defect rate is constructed, and a layered failure early warning threshold is set to automatically generate a secondary repair scheme through a digital twin. Meanwhile, a layered data closed-loop model of detection-repair-review is established to continuously optimize detection parameters through deep learning. The application realizes whole-cycle closed-loop management of bridge concrete disease from accurate identification to intelligent repair to dynamic review, and improves repair efficiency and long-term structural safety.

Description

Technical Field

[0001] This invention belongs to the field of bridge concrete disease detection and repair technology, and relates to a method for layered detection of bridge concrete diseases and verification of repair effects. Background Technology

[0002] With the rapid development of transportation infrastructure construction, the safety and durability assessment of bridge structures has become a key task in the field of transportation maintenance. As the fundamental material supporting the overall structural stability, the evolution of internal defects in bridge concrete directly affects the service life of the bridge. Precise non-destructive testing and targeted reinforcement and repair are core means to ensure bridge operational safety and reduce long-term maintenance costs.

[0003] The technology for detecting and verifying the repair effects of layered defects in bridge concrete has become an important approach to achieving refined maintenance. This technology aims to capture the defect status of different layers within the concrete in real time using advanced detection methods, guide the precise filling of high-performance repair materials based on the detection results, and then comprehensively verify the structural integrity after repair through layered quantitative indicators.

[0004] However, existing technologies have significant limitations when dealing with complex inspection scenarios during the service life of bridges. On the one hand, traditional detection methods struggle to achieve a synergistic correlation between defect depth localization and physical attribute identification in a single inspection, leading to a logical disconnect between inspection results and repair material selection. On the other hand, in structures with high reinforcement ratios, electromagnetic induction shielding and overlapping spectral characteristics severely interfere with the extraction of layered information, and existing algorithms lack targeted filtering mechanisms under dynamic load interference environments, resulting in highly distorted base layer inspection signals. Furthermore, current verification methods primarily focus on macroscopic qualitative descriptions, lacking a quantitative evaluation system for the internal density and interfacial bonding strength of the repair layer, and failing to establish an effective feedback model between inspection and repair data, making it difficult to achieve intelligent localization and proactive early warning of repair failure areas. Summary of the Invention

[0005] In view of this, in order to solve the problems mentioned in the background technology, a method for layered detection of bridge concrete defects and verification of repair effects is proposed.

[0006] The objective of this invention can be achieved through the following technical solution: This invention provides a method for detecting and verifying the repair effect of bridge concrete defects by layer, including: obtaining the reflection spectral characteristics and steel reinforcement distribution of different depth layers inside the bridge concrete through a dual-channel mutual verification and positioning mechanism, and analyzing the layer location of the defects and the corresponding physical property type.

[0007] The filler material system is matched based on the physical property type, and the ratio parameters of the repair material are adjusted according to the environmental conditions at the depth of the lesion.

[0008] A layered correction model for steel reinforcement shielding was constructed to correct the original spectral characteristics layer by layer, and the spectral characteristics of the actual defects that are not affected by steel reinforcement shielding were analyzed.

[0009] Establish a stratified quantitative evaluation index system for the effect of bridge concrete repair.

[0010] Based on the differences in the sensitivity of different concrete depth layers to vibration response, filter parameter sets for each depth layer are configured to eliminate the distortion of detection signals caused by dynamic loads.

[0011] Set a tiered failure warning threshold. When each evaluation indicator in the tiered quantitative evaluation index system exceeds the warning threshold, identify and locate the tiered failure area, and call the historical repair case library to generate a secondary repair plan.

[0012] A mapping relationship between hierarchical detection features and remediation effects for different disease types was established, and remediation parameters were dynamically optimized.

[0013] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. The present invention achieves synchronous and accurate identification of disease depth and attributes through a dual-channel mutual verification mechanism, solving the problem of disconnect between detection and repair in traditional methods.

[0014] 2. This invention improves the reliability of layered detection of highly reinforced bridges under service conditions by using steel bar shielding correction and dynamic load filtering algorithms.

[0015] 3. The hierarchical quantitative evaluation system constructed in this invention transforms the repair effect from a macroscopic qualitative description into a quantitative indicator at a deeper level, providing a scientific basis for refined management and maintenance.

[0016] 4. This invention enables proactive intervention in the repair process by setting up a layered failure early warning mechanism and a digital twin. The introduction of the detection-verification data closed-loop model forms a continuously evolving intelligent detection paradigm, effectively improving the overall efficiency and long-term safety of bridge concrete defect treatment. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a schematic diagram of the method steps of the present invention.

[0019] Figure 2 This is a schematic diagram of the actual disease spectral feature analysis process corresponding to the embodiments of the present invention.

[0020] Figure 3 This is a schematic diagram illustrating the process of configuring the filter parameter set for each depth layer according to an embodiment of the present invention. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] Please see Figure 1 As shown, this invention provides a method for layered detection of bridge concrete defects and verification of repair effects. The specific steps are as follows:

[0023] By using a dual-channel mutual verification positioning mechanism, the reflectance spectral characteristics and steel reinforcement distribution of different depth layers inside the bridge concrete are obtained, and the layer location and corresponding physical property type of the defects are analyzed.

[0024] It's important to explain why we obtain the reflectance spectral characteristics of different depth layers within the bridge concrete and the distribution of reinforcing steel: the reflectance spectral characteristics contain information on energy absorption peak shifts caused by internal defects, which can preliminarily identify the physical properties of defects, such as cracks and voids; the distribution of reinforcing steel can cause signal shielding effects, directly interfering with the accuracy of defect detection. Combining these two methods allows for feature point matching and mapping to a 3D mesh model, precisely locating the layered location of defects. This solves the problem of traditional methods struggling to simultaneously obtain defect depth and properties, providing a core basis for subsequent matching of repair materials and development of appropriate solutions, ensuring the correlation and accuracy of detection and repair.

[0025] It should be noted that the purpose of analyzing the layered location and corresponding physical property type of the defects is as follows: the layered location can clearly define the specific distribution of the defects in the depth direction, avoiding material waste or reinforcement failure caused by indiscriminate identification of depth during repair; the physical property type can directly match suitable filler systems. Combined with the detection data from the dual-channel cross-verification mechanism, this analysis can overcome the limitation of the disconnect between depth and property identification in traditional detection, providing a core basis for subsequent adjustment of repair material ratios and the development of targeted solutions, ensuring that the repair accurately matches the actual condition of the defects, and improving the repair effect and the long-term safety of the bridge structure.

[0026] In a preferred embodiment of the present invention, the specific analysis steps for analyzing the layered location and corresponding physical attribute type of the disease are as follows: a time synchronization trigger is used to control the hyperspectral imaging unit and the electromagnetic induction probe to synchronously collect data on the same scanning path. The dual-channel mutual verification positioning mechanism includes a first channel and a second channel, wherein the first channel collects the reflection spectral characteristics of different depth layers based on hyperspectral imaging technology, and the second channel detects the shielding effect of the steel reinforcement distribution on the disease signal based on electromagnetic induction technology.

[0027] The hyperspectral imaging unit acquires reflectance spectral data within a preset spectral response range, and the spectral data includes information on the shift of energy absorption peaks due to internal defects.

[0028] It needs to be explained that defects in different physical properties inside concrete will cause differential absorption of spectral energy at specific wavelengths, resulting in the position and intensity of the energy absorption peak in the spectral curve deviating from the baseline state of normal concrete.

[0029] Furthermore, the energy absorption peak shift is calculated as follows: using defect-free concrete as a benchmark, its reflectance spectrum data within the preset spectral response range is collected to determine the standard wavelength position of the energy absorption peak under normal conditions; then, reflectance spectrum data of concrete containing internal defects is collected to locate the actual wavelength position of the absorption peak corresponding to the defect; by calculating the difference between the actual wavelength and the standard wavelength, the energy absorption peak shift is obtained, which can reflect the defect type and severity.

[0030] The electromagnetic induction probe generates a secondary magnetic field inside the concrete by emitting electromagnetic waves of a preset frequency, and detects the magnetic field distribution distortion caused by internal voids, cracks or corrosion.

[0031] It needs to be explained that magnetic field distribution distortion is the core phenomenon of electromagnetic induction technology in detecting bridge concrete defects: the electromagnetic induction probe emits electromagnetic waves of a preset frequency, forming a uniform secondary magnetic field inside the defect-free concrete; when encountering defects such as voids, cracks, and corrosion, the internal medium properties of the concrete undergo abrupt changes, causing the originally uniform magnetic field to exhibit an abnormal state of uneven intensity and trajectory deviation.

[0032] Furthermore, the method for calculating magnetic field distribution distortion is as follows: magnetic field data of defect-free concrete is collected as a standard to establish a benchmark distribution model of uniform secondary magnetic field; actual magnetic field data of defective concrete is obtained through an electromagnetic induction probe, and the two are mapped to the same three-dimensional mesh voxel unit; the deviation between the actual magnetic field strength of each voxel and the benchmark value is calculated to quantify the degree of magnetic field strength unevenness and trajectory deviation, so as to accurately reflect the distortion caused by defects such as voids and cracks, and provide data support for defect judgment.

[0033] The spectral image data of the first channel and the electromagnetic intensity distribution data of the second channel are mapped to a unified three-dimensional mesh model by a feature point matching algorithm. The three-dimensional mesh model is composed of voxel units of a preset size.

[0034] The physical property type of the diseased area is initially identified by using the reflectance spectral features obtained by hyperspectral imaging, and the precise location of the diseased area in the concrete depth direction is determined by combining the abnormal changes in dielectric constant fed back by electromagnetic induction data. This enables a joint determination of the layered location of the disease in the concrete structure and its corresponding physical property type.

[0035] It should be added that the method of using the reflectance spectral features obtained by hyperspectral imaging to initially identify the physical property type of the diseased area is as follows: by using a hyperspectral imaging unit to collect reflectance spectral data of different depth layers inside the bridge concrete within a preset spectral response range, the data contains information on the shift of energy absorption peaks caused by internal defects; then the collected spectral data is compared with the reference spectrum of defect-free concrete, and by analyzing the differences in characteristics such as the position shift of absorption peaks, intensity changes, and peak broadening, the physical property type of the disease can be initially distinguished.

[0036] For example, using the reflectance spectrum of defect-free concrete as a benchmark, spectral data of the affected area within a preset spectral response range are collected. The attributes are then distinguished by comparing differences in core characteristics: microcracks significantly increase the absorption intensity of specific wavelengths, with no significant shift in the absorption peak position; voids cause broadening of the absorption peak, with the peak value shifting towards longer wavelengths, and an overall weakening of the absorption intensity; and steel corrosion exhibits unique characteristic absorption peaks. By utilizing these differences in the positional shift, intensity variation, and peak shape of the absorption peaks, it is possible to directly and preliminarily determine whether the defect is a crack, void, or corrosion-related defect.

[0037] The filler material system is matched based on the physical property type, and the ratio parameters of the repair material are adjusted according to the environmental conditions at the depth of the lesion.

[0038] In a preferred embodiment of the present invention, the specific method for matching the filler material system and adjusting the ratio parameters of the repair material is as follows: a structured database containing epoxy resin-based, polyurethane-based and cement-based main materials is pre-constructed, and each type of main material corresponds to a sub-formulation set for different environmental conditions.

[0039] When the type of damage is a micro-crack within a preset size, use a low-viscosity epoxy resin-based material.

[0040] When the type of damage is large-area peeling or voids exceeding a preset depth, use high-flow polyurethane-based materials or cement-based materials with shrinkage compensation properties.

[0041] Real-time humidity, ambient temperature, and historical load data of the disease stratification sites are obtained, and the proportion of hydrophilic functional groups in the curing agent or the proportion of waterproof and water-reducing agent added in the repair material are adjusted according to the humidity data.

[0042] It should be noted that the adjustment of the proportion of hydrophilic functional groups in the curing agent or the addition ratio of waterproofing and water-reducing agent in the repair material should be based on the real-time humidity of the stratification site, combined with environmental temperature and historical load data: When the humidity is high, increase the proportion of hydrophilic functional groups in the curing agent to improve the wetting compatibility of the repair material with the damp substrate, and at the same time, increase the addition ratio of waterproofing and water-reducing agent to enhance the material's water resistance and moisture-proof performance; when the humidity is low, appropriately reduce the proportion of hydrophilic functional groups to avoid shrinkage and cracking of the material due to the dry environment, or reduce the amount of waterproofing and water-reducing agent to ensure the interfacial adhesion between the repair layer and the original concrete. The entire adjustment process must be constrained by the material's fluidity and focused on durability to ensure suitability to the stratification environment and load requirements.

[0043] Furthermore, the determination of the proportion of hydrophilic functional groups in the curing agent or the proportion of waterproof and water-reducing agent in the repair material needs to be combined with a preset structured database and simulation analysis of the layer environment and load data: First, query the basic mix ratio of the corresponding main material in the database under similar temperature, humidity and load conditions; then, by simulating the carbonization depth and chloride ion permeability within a preset period of the layer depth, with durability as the core indicator and fluidity as the constraint, reversely calculate the increase or decrease of hydrophilic functional groups or waterproof and water-reducing agents to ensure that the adjusted material is suitable for the layer environment and meets the repair needs.

[0044] Using the fluidity index of the filler as a constraint and the durability index as the core evaluation parameter, the mixing ratio scheme of the repair material is calculated by simulating the carbonization depth and chloride ion permeability of the layer at this depth within a preset period.

[0045] It should be noted that, firstly, the main materials are selected based on the type of damage, and the basic mix proportions under similar conditions are obtained. Then, the carbonization depth and chloride ion permeability within the preset cycle of the stratum are simulated to assess whether the durability meets the standards, while simultaneously testing whether the material flowability meets the constraints. If the durability is insufficient, the proportion of hydrophilic functional groups in the curing agent and the amount of waterproofing and water-reducing agent are increased if the humidity is high, and the opposite is true if the humidity is low. If the flowability does not meet the standards, the aggregate gradation or the amount of admixtures can be fine-tuned. The simulation and parameter adjustment are iterated repeatedly until both core indicators are suitable for the stratum requirements, and the final mix proportion scheme is determined.

[0046] Furthermore, to determine whether the fluidity of the repair material meets the standard, its slump, spread, and other indicators need to be measured and compared with the preset fluidity threshold for the depth of the damaged layer. If the measured value is not lower than the threshold, it meets the standard. The durability standard is determined by simulating the carbonation depth and chloride ion permeability within a preset period at the depth layer. The simulation results are compared with the durability safety threshold of the bridge concrete structure. Only if both indicators meet the specified requirements can it be determined that it meets the standard.

[0047] A layered correction model for steel reinforcement shielding was constructed to correct the original spectral characteristics layer by layer, and the spectral characteristics of the actual defects that are not affected by steel reinforcement shielding were analyzed.

[0048] It should be noted that the reason for constructing the rebar shielding layer correction model is that the rebar inside the concrete can significantly shield and interfere with electromagnetically induced magnetic field signals and hyperspectral reflectance spectral signals, easily leading to misalignment of the defect layer location and distortion of physical property identification, directly affecting the accuracy of the detection data. This model can specifically quantify the degree of shielding interference from the rebar, perform layer correction and deviation correction on the interfered detection data, restore the true depth coordinates and attribute characteristics of the defect, and provide reliable corrected data support for subsequent accurate defect identification and repair plan development.

[0049] For a preferred embodiment of the present invention, please refer to Figure 2 As shown, the specific analysis method for constructing the rebar shielding layer correction model is as follows: a multilayer perceptron neural network architecture is adopted, which includes an input layer, multiple hidden layers and an output layer. The feature vector of the input layer is the gradient change sequence of electromagnetic induction intensity in the direction of concrete depth. By learning the signal distribution law of known rebar arrangement samples, the spatial arrangement density and geometric contour of the rebar can be inverted.

[0050] Calculate the spectral correction factor for each depth layer. The spectral correction factor is equal to the ratio of the actual reflectance intensity to the observed reflectance intensity. The value of the spectral correction factor is determined by the steel reinforcement density, the detection depth, and the attenuation coefficient of the concrete material.

[0051] The spectral correction factor is multiplied pixel by pixel with the original hyperspectral data sequence to correct the signal attenuation deviation caused by the steel bar occlusion in the original hyperspectral data, thereby extracting the true spectral features of the disease that are not affected by the steel bar shielding.

[0052] Establish a stratified quantitative evaluation index system for the effect of bridge concrete repair.

[0053] In a preferred embodiment of the present invention, the specific construction method of the layered quantitative evaluation index system is as follows: the layered quantitative evaluation index system includes the uniformity of density layering inside the repair layer, the layered gradient of the bond strength between the repair layer and the original concrete interface, and the layered residual defect rate after the repair of the original concrete disease area. Each index is divided into multiple evaluation layers according to the concrete depth direction and is assigned a corresponding numerical scoring weight.

[0054] It should be noted that the above three indicators are included in the stratified quantitative evaluation indicator system because they accurately reflect the actual quality of bridge concrete layered repair from different core dimensions and are suitable for the needs of depth layered testing: the uniformity of density layering determines the structural stability of the repair layer itself; the layered gradient of interfacial bond strength ensures the bonding force between the repair layer and the original concrete layer and avoids interlayer delamination; and the layered residual defect rate directly verifies the effectiveness of the repair and radical treatment of the diseased area. The stratified quantification and weighting of these three indicators can comprehensively and accurately assess the repair compliance of each depth layer, providing a scientific quantitative basis for repair quality acceptance and subsequent maintenance.

[0055] It should be noted that the numerical scoring weights should be determined based on the disease risk level and structural stress characteristics of each depth assessment layer: for depth layers with concentrated disease and high structural stress, the scoring weights of the three indicators are increased, with the layer residual defect rate, which plays a core role in the repair effect, having the highest weight; for depth layers with minor disease and non-core stress, the overall weight is appropriately reduced, and the interfacial bonding strength layer gradient and density uniformity are assigned secondary weights according to the functional requirements of the layer. All layer weights are normalized to ensure that the total weight is 1.

[0056] The ultrasonic transmission method was used to measure layer by layer in the direction perpendicular to the repair surface. The ultrasonic transducer was used to step along the depth direction, and the propagation time and sound velocity dispersion of the sound wave in each layer were recorded. The sound velocity dispersion was used to characterize the density and layer uniformity of the repair layer.

[0057] It should be explained that sound velocity dispersion is a quantitative indicator in ultrasonic testing of concrete, characterizing the degree of dispersion of sound wave propagation velocity at various measuring points within the tested area. It primarily reflects the uniformity of the internal structure of the concrete and is a key parameter for layered testing of bridge concrete. The calculation involves first obtaining the actual sound velocity values ​​at multiple measuring points within the tested layer, calculating their arithmetic mean and standard deviation, and then using the ratio of the standard deviation to the mean to obtain the dispersion value. A larger dispersion indicates greater unevenness in the internal density of the concrete, indirectly confirming the existence or distribution differences of defects. It is often used to assist in assessing the uniformity of the density of repair layers.

[0058] By applying shear or tensile forces at different depth sections through micro pull-out tests and recording the load values ​​at material failure, the layered gradient of the bond strength between the repair layer and the original concrete interface is obtained. If the bond strength values ​​at each depth fluctuate within a preset ratio, the layered gradient is deemed qualified.

[0059] It should be noted that the preset ratio is based on the design code for bridge concrete structures, combined with the bulk bond strength of the original concrete substrate, the mechanical performance parameters of the repair material, and reference to the industry's general standards for allowable fluctuations in interfacial bond strength. Simultaneously, it takes into account the structural stress differences at different depths of the damage layers, the interlayer anti-peeling requirements for long-term bridge service, and calibration using micro-pull-out test data from similar projects. Finally, a suitable preset ratio for numerical fluctuation is determined to ensure that the judgment criteria align with the actual engineering requirements and structural safety requirements.

[0060] Image semantic segmentation technology was used to normalize the hyperspectral cube data before and after repair and to annotate the pixels in the diseased area. The percentage of pixels annotated as diseased after repair was used as the percentage of the total number of pixels in the original diseased area, and this percentage was used as the layered residual defect rate.

[0061] Based on the differences in the sensitivity of different concrete depth layers to vibration response, filter parameter sets for each depth layer are configured to eliminate the distortion of detection signals caused by dynamic loads.

[0062] For a preferred embodiment of the present invention, please refer to Figure 3 As shown, the specific method for eliminating the distortion of detection signals caused by dynamic load is as follows: the vibration frequency and amplitude signals of the bridge under vehicle traffic or wind vibration are collected in real time by a high-frequency acceleration sensor.

[0063] It should be explained that because bridges experience continuous, minute vibrations due to vehicle traffic or wind-induced vibrations, these dynamic loads cause a significant amount of random noise to be mixed into the base layer signals acquired by the detection sensors. This algorithm first uses a high-frequency accelerometer to collect the bridge's vibration frequency and amplitude signals in real time.

[0064] Furthermore, the algorithm employs an adaptive wavelet packet decomposition structure to perform multi-scale decomposition of the acquired detection signals in the time-frequency domain. Since the sensitivity of concrete layers at different depths to vibration response varies significantly—shallow layers are greatly affected by high-frequency vibrations, while deeper layers are significantly affected by low-frequency structural vibrations—the algorithm configures a dedicated set of filtering parameters for each depth layer. Specifically, the system dynamically calculates the suppression coefficient based on the frequency sub-band position of the current dominant bridge vibration frequency. During wavelet packet reconstruction, the system automatically reduces the weight of signal components that coincide with the dynamic load frequency, while simultaneously enhancing the frequency band containing defect feature information through an energy compensation mechanism. This deep-layer-specific filtering eliminates the distortion of the base layer detection signal caused by dynamic loads, significantly improving the detection accuracy of the bridge without disrupting traffic.

[0065] An adaptive wavelet packet decomposition structure is used to decompose the acquired detection signal into multiple scales in the time and frequency domain, and the suppression coefficient is dynamically calculated based on the frequency sub-band position of the bridge vibration main frequency.

[0066] It should be noted that, firstly, wavelet packet decomposition is used to extract each frequency sub-band of the detection signal and calculate the sub-band energy to locate the main frequency sub-band of bridge vibration; then, based on the energy of the main frequency sub-band, the energy ratio of each interference sub-band to it is calculated and substituted into the preset suppression coefficient calculation formula; finally, combined with the actual signal-to-noise ratio of the detection signal and the dynamic correction result, the final suppression coefficient of each interference sub-band is determined, so as to achieve accurate suppression of non-main frequency interference.

[0067] Furthermore, the formula for calculating the inhibition coefficient is as follows: ,in For the first The basic suppression coefficient of each interference frequency sub-band For the first Signal energy of each interference frequency sub-band This represents the signal energy of the sub-band corresponding to the dominant frequency of bridge vibration. It is a very small positive value, and its range is [value range missing]. To avoid the denominator being 0.

[0068] When performing wavelet packet reconstruction, the weight of signal components that coincide with the dynamic load frequency is reduced, and the frequency band containing the fault characteristic information is enhanced through the energy compensation mechanism, thereby eliminating the distortion of the base detection signal caused by the dynamic load.

[0069] It should be noted that in the wavelet packet reconstruction stage, the signal sub-band that coincides with the dynamic load frequency is first accurately located, and the basic suppression coefficient corresponding to the sub-band is calculated by substituting it into the suppression coefficient formula. The signal amplitude of this sub-band is multiplied by (1-suppression coefficient) to complete the weight reduction. The reduction range is determined by the suppression coefficient. The higher the energy proportion of the dynamic load sub-band, the larger the suppression coefficient and the more weight reduction. At the same time as the reduction, the energy compensation mechanism is activated to ensure that the signal energy of the defective frequency band is not lost.

[0070] Set a tiered failure warning threshold. When each evaluation indicator in the tiered quantitative evaluation index system exceeds the warning threshold, identify and locate the tiered failure area, and call the historical repair case library to generate a secondary repair plan.

[0071] In a preferred embodiment of the present invention, the specific analysis method for identifying and locating the failure layered region and generating a secondary repair scheme is as follows: a digital twin is established, which stores the layered state data of the bridge throughout its entire life cycle based on a three-dimensional spatial index structure, and the real-time acquired verification data is compared with preset density thresholds, interface bonding strength thresholds, and residual defect rate thresholds.

[0072] It should be noted that the threshold values ​​for density, interfacial bond strength, and residual defect rate are set based on the design specifications for bridge concrete structures and industry standards for repair quality acceptance. They are determined by combining the mechanical properties of the original concrete substrate, the inherent characteristics of the repair materials, and the stress differences in each layer of the bridge structure. At the same time, they are calibrated by relying on the layered state data of the bridge throughout its entire life cycle in the digital twin and by referring to the actual test data of similar projects. The threshold values ​​take into account the durability of the bridge in long-term service, the resistance to interlayer peeling, and the need for the complete eradication of defects, so that the threshold values ​​are consistent with the actual project and meet the requirements for repair quality judgment and structural safety assurance.

[0073] When the measurement data of a certain layer exceeds the warning threshold, the digital twin activates the geometric topology model and material property data of the failure area.

[0074] The case-based reasoning algorithm searches the historical repair case database for the record with the highest similarity to the current failure characteristics, and generates a secondary repair instruction set that includes material ratio suggestions, injection pressure settings, and maintenance cycle parameters.

[0075] It should be explained that the highest similarity to the current failure characteristics refers to the optimal match between the core characteristics of the current bridge concrete repair failure and the failure characteristics of cases in the historical repair case database. These core characteristics include defect type, delamination location, density, and degree of bond strength deficiency. In the quantitative analysis, the characteristic indicators of both sides are first standardized, and then the cosine similarity algorithm is used to calculate the similarity value between the features. The closer the value is to 1, the higher the similarity. The historical case corresponding to the maximum similarity within the threshold is selected as the target case.

[0076] A mapping relationship between hierarchical detection features and remediation effects for different disease types was established, and remediation parameters were dynamically optimized.

[0077] In a preferred embodiment of the present invention, the dynamic optimization of repair parameters is carried out as follows: a graph neural network architecture is adopted, and each detection, repair and review task is modeled as a graph node. The attribute set of the node includes the disease type, layer depth distribution, repair scheme parameters and review evaluation results.

[0078] The edge weights between nodes are set according to the technical similarity or geographical correlation between different tasks, and the nonlinear mapping relationship between the disease evolution pattern and the repair response is explored through multiple rounds of iterative learning in the graph via message passing mechanism.

[0079] The repair parameters are dynamically optimized based on the learning results, and the weight values ​​on the instruction paths that successfully improve detection accuracy or repair pass rate are increased according to the reward and punishment mechanism updated by edge weights.

[0080] It's important to note that the edge weight updates in graph neural networks follow a reward-penalty mechanism. If a feedback instruction successfully improves the signal-to-noise ratio of subsequent detections or the pass rate of repairs, the system increases the weight value on the path that generated that instruction; conversely, it decreases it. Through this reinforcement learning-style evolution, the system can identify extremely subtle correlation features. For example, there is a leading correlation between anomalous electromagnetic wave signals of a specific frequency and the evolution of deep microcracks in concrete. Once this pattern is solidified into the initial detection model, the system can achieve advanced prediction before the crack width reaches the millimeter level, using only weak spectral shifts and magnetic field gradient perturbations.

[0081] In a preferred embodiment of the present invention, the method further includes an environmental adaptation step, the specific process of which is as follows: when the detection environment is an underwater environment, a sealed chamber with active light source compensation is used to offset the reduction of spectral energy by water, and a coil with low-frequency penetration characteristics is used to reduce the interference of water medium conductivity on the magnetic field.

[0082] When adjusting the ratio of repair materials, increase the underwater non-dispersant component and increase the proportion of coagulant according to the decrease in underwater temperature.

[0083] A medium correction factor is introduced into the calibration model. The input characteristics of the medium correction factor include conductivity and dielectric constant, which is used to isolate the influence of the water medium from the steel reinforcement shielding effect.

[0084] It is important to note that the core of environmental adaptation is to address the significant interference of the unique underwater testing environment on the detection signals of concrete defects, thus preventing signal distortion from affecting detection accuracy. Water absorbs and scatters spectral energy, causing attenuation, and the electrical conductivity of the water medium also interferes with the magnetic field signal. Both of these factors lead to inaccurate detection data and an inability to accurately identify concrete defects. By employing targeted environmental adaptation methods to counteract these interferences, the stability and accuracy of spectral and magnetic field detection signals can be guaranteed, ensuring the reliability of underwater concrete defect detection.

[0085] In a preferred embodiment of the present invention, the method further includes preventive maintenance alarm analysis, the specific process of which is as follows: using a long short-term memory network to analyze the trend of historical review data, predicting the evolution path of various evaluation indicators within a preset time period in the future, and issuing a preventive maintenance alarm before the predicted indicators fall below the warning threshold.

[0086] It should be noted that the core advantage of this preventive maintenance alarm analysis process lies in its ability to accurately predict the evolution path of assessment indicators based on historical review data, leveraging the time-series data processing advantages of long short-term memory networks. This enables early warning of disease risks, breaking through the passive mode of traditional post-maintenance. By issuing alarms before indicators fall below warning thresholds, maintenance work can be carried out in advance, preventing further deterioration of diseases and significantly reducing the cost of later major repairs. Simultaneously, it proactively manages bridge structural safety risks, enhancing the scientific and forward-looking nature of maintenance work and effectively extending the overall service life of the bridge.

[0087] The above content is merely an example and illustration of the concept of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described, or use similar methods to replace them, as long as they do not deviate from the concept of the invention or exceed the scope defined by the present invention, and all such modifications and additions should fall within the protection scope of the present invention.

Claims

1. A method for layered detection of bridge concrete defects and verification of repair effectiveness, characterized in that: include: By using a dual-channel mutual verification positioning mechanism, the reflectance spectral characteristics and steel reinforcement distribution of different depth layers inside the bridge concrete are obtained, and the layer location and corresponding physical property type of the defects are analyzed. The specific analytical steps for analyzing the stratification location and corresponding physical attribute types of the disease are as follows: A time-synchronization trigger is used to control the hyperspectral imaging unit and the electromagnetic induction probe to synchronously acquire data on the same scanning path. The dual-channel mutual verification positioning mechanism includes a first channel and a second channel. The first channel acquires the reflection spectrum characteristics of different depth layers based on hyperspectral imaging technology, and the second channel detects the shielding effect of the steel reinforcement distribution on the defect signal based on electromagnetic induction technology. The hyperspectral imaging unit acquires reflectance spectral data within a preset spectral response range, and the spectral data includes information on the shift of energy absorption peaks due to internal defects. The electromagnetic induction probe generates a secondary magnetic field inside the concrete by emitting electromagnetic waves of a preset frequency, and detects the magnetic field distribution distortion caused by internal voids, cracks or corrosion. The spectral image data of the first channel and the electromagnetic intensity distribution data of the second channel are mapped to a unified three-dimensional mesh model by a feature point matching algorithm. The three-dimensional mesh model is composed of voxel units of a preset size. The physical property type of the diseased area is initially identified by using the reflectance spectral features obtained by hyperspectral imaging, and the precise location of the diseased area in the concrete depth direction is determined by combining the abnormal changes in dielectric constant fed back by electromagnetic induction data. This enables a joint determination of the layered location of the disease in the concrete structure and its corresponding physical property type. The filler material system is matched based on the physical property type, and the ratio parameters of the repair material are adjusted according to the environmental conditions at the depth of the disease. A layered correction model for steel reinforcement shielding was constructed to correct the original spectral characteristics layer by layer, and the spectral characteristics of the actual defects that are not affected by steel reinforcement shielding were analyzed. The specific analysis method for constructing the rebar shielding layered correction model is as follows: A multilayer perceptron neural network architecture is adopted, which includes an input layer, multiple hidden layers and an output layer. The feature vector of the input layer is the gradient change sequence of electromagnetic induction intensity in the direction of concrete depth. By learning the signal distribution law of known steel bar arrangement samples, the spatial arrangement density and geometric contour of the steel bars can be inverted. Calculate the spectral correction factor for each depth layer. The spectral correction factor is equal to the ratio of the actual reflectance intensity to the observed reflectance intensity. The value of the spectral correction factor is determined by the steel reinforcement density, the detection depth, and the attenuation coefficient of the concrete material. The spectral correction factor is multiplied pixel by pixel with the original hyperspectral data sequence to correct the signal attenuation deviation caused by steel bar occlusion in the original hyperspectral data, thereby extracting the true spectral features of the disease that are not affected by steel bar shielding. Establish a hierarchical and quantitative evaluation index system for the effect of bridge concrete repair; The specific construction method of the hierarchical quantitative evaluation index system is as follows: The layered quantitative evaluation index system includes the uniformity of density layering inside the repair layer, the layered gradient of the bond strength between the repair layer and the original concrete interface, and the layered residual defect rate after the repair of the original concrete diseased area. Each index is divided into multiple evaluation layers according to the concrete depth direction and is assigned a corresponding numerical scoring weight. The ultrasonic transmission method was used to measure layer by layer in the direction perpendicular to the repair surface. The ultrasonic transducer was used to step along the depth direction and the propagation time and sound velocity dispersion of the sound wave in each layer were recorded. The sound velocity dispersion was used to characterize the density and layer uniformity of the repair layer. By applying shear or tensile force to cross sections at different depths through micro pull-out tests and recording the load values ​​at material failure, the layer gradient of the bond strength between the repair layer and the original concrete interface is obtained. If the bond strength values ​​at each depth fluctuate within a preset ratio, the layer gradient is deemed qualified. Image semantic segmentation technology was used to normalize the hyperspectral cube data before and after repair and to label the pixels in the diseased area. The percentage of pixels labeled as diseased after repair was counted to the total number of pixels in the original diseased area, and this percentage was used as the layered residual defect rate. Based on the differences in the sensitivity of different concrete depth layers to vibration response, filter parameter sets for each depth layer are configured to eliminate the distortion of detection signals caused by dynamic loads. Set a tiered failure warning threshold. When each evaluation indicator in the tiered quantitative evaluation index system exceeds the warning threshold, identify and locate the tiered failure area, and call the historical repair case library to generate a secondary repair plan. A mapping relationship between hierarchical detection features and remediation effects for different disease types was established, and remediation parameters were dynamically optimized.

2. The method for layered detection and verification of repair effects of bridge concrete defects according to claim 1, characterized in that: The specific methods for matching the filler material system and adjusting the ratio parameters of the repair material are as follows: A structured database containing epoxy resin-based, polyurethane-based and cement-based main materials is pre-constructed, with each type of main material corresponding to a sub-formulation set for different environmental conditions; When the type of damage is a micro-crack within a preset size, use a low-viscosity epoxy resin-based material. When the type of disease is large-area peeling or voids exceeding a preset depth, use high-flow polyurethane-based materials or cement-based materials with shrinkage compensation properties. Obtain real-time humidity, ambient temperature and historical load data of the disease stratification sites, and adjust the proportion of hydrophilic functional groups of the curing agent or the proportion of waterproof and water-reducing agent added in the repair material according to the humidity data. Using the fluidity index of the filler as a constraint and the durability index as the core evaluation parameter, the mixing ratio scheme of the repair material is calculated by simulating the carbonization depth and chloride ion permeability of the layer at this depth within a preset period.

3. The method for layered detection and verification of repair effects of bridge concrete defects according to claim 1, characterized in that: The specific method for eliminating detection signal distortion caused by dynamic load is as follows: The vibration frequency and amplitude signals of the bridge under vehicle traffic or wind vibration are collected in real time using a high-frequency accelerometer. An adaptive wavelet packet decomposition structure is used to decompose the acquired detection signal in the time and frequency domain at multiple scales, and the suppression coefficient is dynamically calculated based on the frequency sub-band position of the bridge vibration main frequency. When performing wavelet packet reconstruction, the weight of signal components that coincide with the dynamic load frequency is reduced, and the frequency band containing the fault characteristic information is enhanced through the energy compensation mechanism, thereby eliminating the distortion of the base detection signal caused by the dynamic load.

4. The method for layered detection and verification of repair effects of bridge concrete defects according to claim 3, characterized in that: The specific analysis method for identifying and locating the failure layered region and generating a secondary repair plan is as follows: A digital twin is established, which stores the layered status data of the bridge throughout its entire life cycle based on a three-dimensional spatial index structure, and compares the real-time acquired verification data with preset density thresholds, interface bonding strength thresholds, and residual defect rate thresholds. When the measurement data of a certain layer exceeds the warning threshold, the digital twin activates the geometric topology model and material property data of the failure area; The case-based reasoning algorithm searches the historical repair case database for the record with the highest similarity to the current failure characteristics, and generates a secondary repair instruction set that includes material ratio suggestions, injection pressure settings, and maintenance cycle parameters.

5. The method for layered detection and verification of repair effects of bridge concrete defects according to claim 1, characterized in that: The method for dynamically optimizing and repairing parameters is as follows: A graph neural network architecture is adopted to model each detection, repair and review task as a graph node. The attribute set of the node includes the disease type, layer depth distribution, repair scheme parameters and review evaluation results. The edge weights between nodes are set according to the technical similarity or geographical correlation between different tasks, and the nonlinear mapping relationship between the disease evolution pattern and the repair response is explored through multiple rounds of iterative learning in the graph via message passing mechanism. Based on the message passing mechanism, the repair parameters are dynamically optimized through multiple rounds of iterative learning in the graph. The weight values ​​on the instruction paths that successfully improve detection accuracy or repair pass rate are increased according to the reward and punishment mechanism of edge weight updates.

6. The method for layered detection and verification of repair effects of bridge concrete defects according to claim 1, characterized in that: The method also includes an environment adaptation step, the specific process of which is as follows: When the detection environment is underwater, the reduction of spectral energy by water is offset by a sealed chamber with active light source compensation, and the interference of water conductivity on magnetic field is reduced by a coil with low frequency penetration characteristics. When adjusting the ratio of repair materials, increase the underwater non-dispersant component and increase the proportion of coagulant according to the decrease in underwater temperature; A medium correction factor is introduced into the calibration model. The input characteristics of the medium correction factor include conductivity and dielectric constant, which is used to isolate the influence of the water medium from the steel reinforcement shielding effect.

7. The method for layered detection and verification of repair effects of bridge concrete defects according to claim 1, characterized in that: The method also includes preventative maintenance alarm analysis, the specific process of which is as follows: By using long short-term memory networks to analyze the trends of historical review data, the evolution path of various assessment indicators within a preset time period can be predicted, and a preventive maintenance alarm can be issued before the predicted indicators fall below the warning threshold.

Images