A method and device for precise identification and localization of bridge damage based on multi-source data fusion

Abstract

This invention discloses a method and apparatus for accurate bridge damage identification and location based on multi-source data fusion. The method includes: performing spatiotemporal registration and data fusion on the time-series data of the bridge's global planar deformation obtained by ground-based synthetic aperture radar and the time-series data of the bridge's point-like three-dimensional deformation obtained by the BeiDou satellite navigation system to obtain a fused deformation field of the entire bridge area; calculating multiple damage characteristic indicators reflecting different deformation physical mechanisms of the bridge based on the fused deformation field, and determining the corresponding preliminary damage level based on the damage judgment threshold of each damage characteristic indicator; using the preliminary damage levels of each damage characteristic indicator as independent evidence sources, performing evidence fusion to resolve indicator conflicts, and determining the bridge damage location and damage level based on the fusion result. This method can effectively improve the accuracy and reliability of bridge damage identification and location, providing a scientific basis for bridge operation and maintenance decisions.

Description

Technical Field

[0001] This application relates to the field of bridge monitoring technology, specifically to a method and device for accurate identification and location of bridge damage based on multi-source data fusion. Background Technology

[0002] Bridges are an important component of transportation infrastructure, and their structural safety plays a crucial role in ensuring traffic safety. As bridges age, coupled with the long-term effects of heavy traffic loads and extreme weather conditions, bridge structures are prone to various types of damage, including crack propagation, bearing aging, and main structural deformation. Accurately locating the damage and assessing its severity has become a core technical requirement for ensuring the safe operation of bridges and extending their service life.

[0003] Currently, bridge damage detection mainly relies on traditional methods, but these methods have many limitations in practical applications and cannot meet the requirements for precise location. Manual inspection is inefficient and highly subjective, and has limited ability to identify hidden damage; vibration monitoring technology is not sensitive enough to local damage such as microcracks, and the global indicators such as modal parameters it acquires cannot directly reflect the specific location of the damage; although fiber optic sensing technology has high monitoring accuracy, its installation and maintenance process is complex, it is easily affected by construction interference, and its monitoring coverage is limited, making it difficult to achieve full-area monitoring of the bridge.

[0004] In recent years, ground-based synthetic aperture radar (GB-InSAR) technology and the BeiDou Navigation Satellite System (BDS) have been increasingly applied to bridge monitoring. GB-InSAR offers advantages such as non-contact, high precision, wide coverage, and real-time monitoring, effectively acquiring millimeter-level deformation fields of bridge structures. However, it suffers from drawbacks including a single line-of-sight orientation, susceptibility to atmospheric disturbances, and insensitivity to non-deformation damage identification. BDS provides three-dimensional absolute coordinates of key bridge points, offering strong real-time performance and immunity to obstruction. However, single-point monitoring is costly and struggles to achieve complete reconstruction of the bridge's overall deformation field. Furthermore, GB-InSAR and BDS are often used independently, resulting in isolated monitoring data that fails to meet the combined requirements of monitoring range and positioning accuracy.

[0005] Meanwhile, existing damage location methods mostly rely on a single deformation index, leading to insufficient damage location accuracy and misjudgment of damage types, making it difficult to meet the actual needs of precise bridge damage location. Therefore, how to improve the accuracy and reliability of bridge damage identification and location is a technical problem that needs to be solved in this field. Summary of the Invention

[0006] This application provides a method and apparatus for accurate identification and location of bridge damage based on multi-source data fusion, which can solve the technical problems in the prior art of insufficient spatiotemporal coverage of bridge damage monitoring and high misjudgment rate caused by the easy interference of monitoring noise with single damage indicators.

[0007] In a first aspect, embodiments of this application provide a method for accurate identification and location of bridge damage based on multi-source data fusion, the method comprising: The bridge's global planar deformation time-series data were obtained by ground-based synthetic aperture radar, and the bridge's point-like three-dimensional deformation time-series data were obtained by the BeiDou satellite navigation system. The spatiotemporal registration and data fusion of the global planar deformation time series data and the point-like three-dimensional deformation time series data are performed to obtain the fused deformation field of the entire bridge area; Based on the fused deformation field, multiple damage characteristic indicators reflecting different deformation physical mechanisms of the bridge are calculated, and based on the damage judgment threshold of each damage characteristic indicator, the corresponding preliminary damage level is determined. The damage characteristic indicators include the fused coordinate mode confidence criterion, the rate of change of fused internal accumulation, and the deformation gradient anomaly value. The preliminary damage levels of each damage characteristic index are used as independent sources of evidence. Evidence fusion is performed to resolve index conflicts, and the bridge damage location and final damage level are determined based on the fusion results.

[0008] Secondly, embodiments of this application provide a bridge damage precision identification and location device based on multi-source data fusion, the device comprising: The acquisition module is used to acquire the time-series data of the bridge's global planar deformation through ground-based synthetic aperture radar and the time-series data of the bridge's point-like three-dimensional deformation through the BeiDou satellite navigation system. The fusion module is used to perform spatiotemporal registration and data fusion on the global planar deformation time series data and the point-like three-dimensional deformation time series data to obtain the fused deformation field of the entire bridge area. The calculation module is used to calculate multiple damage characteristic indicators reflecting different deformation physical mechanisms of the bridge based on the fused deformation field, and to determine the corresponding preliminary damage level based on the damage judgment threshold of each damage characteristic indicator. The damage characteristic indicators include the fused coordinate mode confidence criterion, the rate of change of fused internal accumulation, and the deformation gradient anomaly value. The determination module is used to use the preliminary damage levels of each damage characteristic index as independent evidence sources, perform evidence fusion to resolve index conflicts, and determine the location and level of bridge damage based on the fusion results.

[0009] The beneficial effects of the technical solutions provided in this application include: The bridge's global planar deformation time-series data were acquired using ground-based synthetic aperture radar, while its point-based three-dimensional deformation time-series data were acquired using the BeiDou Navigation Satellite System. Spatiotemporal registration and data fusion were performed on the global planar deformation time-series data and the point-based three-dimensional deformation time-series data to obtain a fused deformation field covering the entire bridge. Based on the fused deformation field, multiple damage characteristic indicators reflecting different deformation physical mechanisms of the bridge were calculated, and the corresponding preliminary damage level was determined based on the damage judgment threshold of each damage characteristic indicator. These damage characteristic indicators include the fused coordinate mode confidence criterion, the rate of change of fused internal accumulation volume, and deformation gradient anomalies. The preliminary damage levels of each damage characteristic indicator were used as independent evidence sources, and evidence fusion was performed to resolve indicator conflicts. The bridge damage location and damage level were determined based on the fusion results. This approach solves the technical problems in related technologies, such as insufficient spatiotemporal coverage of single monitoring methods, high misjudgment rate of single damage indicators, and large uncertainty in decision-making due to the fusion of multi-source heterogeneous data.

[0010] This application constructs a unified data foundation for the fusion deformation field of multi-source data through spatiotemporal registration; it introduces multi-mechanism damage indicators to provide mutually independent and reliable evidence sources for evidence fusion; and it utilizes evidence theory fusion algorithms to effectively resolve conflicts between indicators and filter out noise interference. Through synergistic effects, these processes jointly overcome the challenges of comprehensive perception, high-precision data fusion, and strong anti-interference judgment in bridge monitoring, ultimately achieving precise comprehensive positioning of bridge damage and reliable assessment of its health status, providing a scientific and objective basis for bridge operation and maintenance decisions. Attached Figure Description

[0011] Figure 1 This is a flowchart illustrating an embodiment of the bridge damage accurate identification and localization method based on multi-source data fusion according to this application. Figure 2 A schematic diagram of the process for generating the fused deformation field for this application; Figure 3 A flowchart illustrating the evidence fusion process for this application; Figure 4 This is a schematic diagram of the functional modules of an embodiment of the bridge damage accurate identification and positioning device based on multi-source data fusion according to this application. Detailed Implementation

[0012] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present application.

[0013] To make the objectives, technical solutions, and advantages of this application clearer, the embodiments of this application will be described in further detail below with reference to the accompanying drawings.

[0014] In a first aspect, embodiments of this application provide a method for accurate identification and location of bridge damage based on multi-source data fusion.

[0015] In one embodiment, reference is made to Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the bridge damage accurate identification and location method based on multi-source data fusion according to this application. Figure 1 As shown, the bridge damage accurate identification and localization method based on multi-source data fusion includes: Step S1: Obtain the time series data of the bridge's global planar deformation using ground-based synthetic aperture radar, and obtain the time series data of the bridge's point-like three-dimensional deformation using the BeiDou satellite navigation system.

[0016] It is worth noting that a ground-based synthetic aperture radar (GB-InSAR) device can be installed on a stable foundation near the bridge to perform non-contact, large-area continuous scanning imaging of the bridge structure. Its output global planar deformation time-series data covers the entire bridge area and is organized in a grid format. The data for each grid cell represents the displacement of the bridge surface along the radar line of sight over time.

[0017] High-precision BeiDou Navigation Satellite System (BDS) terminals can be deployed at key structural points of the bridge, including stress- or deformation-sensitive sections such as mid-span, supports, and pier tops. Using positioning technologies such as carrier phase differential, the three-dimensional coordinates of each monitoring point can be calculated and output in real time.

[0018] Step S2: Perform spatiotemporal registration and data fusion on the global planar deformation time series data and the point-like three-dimensional deformation time series data to obtain the fused deformation field of the entire bridge.

[0019] It is worth noting that in this embodiment, spatiotemporal registration of the global planar deformation time series data and the point-like three-dimensional deformation time series data is performed with unified coordinate system, unified sampling rate, and time synchronization. Kalman filtering is used to construct a displacement-velocity-acceleration three-dimensional state vector. The registered data is then fused and estimated to finally obtain the bridge's global fused deformation field.

[0020] In one embodiment, such as Figure 2 As shown, step S2 specifically includes steps S201 to S204: Step S201: Unify the global planar deformation time series data and the point-like three-dimensional deformation time series data into the geodetic coordinate system.

[0021] Explaining this, WGS84 (World Geodetic System 1984) is a geocentric coordinate system based on the Earth's center of mass. Global planar deformation data output by GB-InSA typically uses the device's local coordinate system or a relative coordinate system, while point-based 3D deformation data output by BDS typically uses the WGS84 geodetic coordinate system. Converting global planar deformation data from the device's local coordinate system to the WGS84 geodetic coordinate system allows for coordinate conversion between the two types of data within the same spatial framework, avoiding spatial deviations caused by coordinate system differences.

[0022] Step S201: Map the point-like three-dimensional deformation time-series data to the monitoring surface element of the ground-based synthetic aperture radar through spatial interpolation, and convert the point-like three-dimensional deformation time-series data into surface grid data with the same spatial resolution as the global surface deformation time-series data.

[0023] Point-based 3D deformation time-series data consists of 3D deformation data from discrete monitoring points (such as monitoring points at key sections of a bridge), while global area deformation time-series data consists of LOS-oriented deformation data in the form of area grids (area elements). The two have inconsistent spatial morphologies and cannot be directly fused. This embodiment uses Kriging interpolation to map the discrete point-based 3D deformation time-series data to GB-InSAR monitoring area elements to achieve unified spatial resolution.

[0024] After establishing coordinate system one, outlier data points in the point-based 3D deformation time-series data are removed to ensure data quality. Based on the spatial distribution characteristics of the point-based 3D deformation time-series data, the corresponding experimental semivariogram is calculated and fitted using a pre-defined theoretical model to obtain a theoretical semivariogram model, which quantifies the spatial deformation correlation. The theoretical model can be a spherical model or an exponential model. The spatial distribution characteristics of the point-based 3D deformation time-series data include the Euclidean distance between monitoring points, azimuth angle, and the distribution of point pairs.

[0025] For each GB-InSAR monitoring cell to be mapped, a set of Kriging equations is constructed based on a theoretical semi-variogram model. The construction of the Kriging equations satisfies the conditions of unbiasedness and minimum estimation variance. The interpolation weights of the monitoring points in each point-based 3D deformation time series data relative to the monitoring cell are solved. Then, the point-based 3D deformation time series data are weighted and summed according to the interpolation weights to calculate the deformation estimate of the corresponding monitoring cell. The above steps are repeated to traverse all monitoring cells, converting the point-based 3D deformation time series data into planar mesh data with the same spatial resolution as the global planar deformation time series data.

[0026] Furthermore, using manual monitoring data at the bridge's fixed supports as a benchmark, deviation correction is applied to the interpolated planar mesh data, with the correction amount controlled within 0.3 mm. Specifically, the average deformation sequence value at the corresponding location of the fixed support is extracted as a systematic deviation, and all planar mesh data undergo overall translation correction. If the deviation exceeds 0.3 mm, the data source is checked and recalculated until the accuracy requirements are met.

[0027] It is worth noting that this embodiment uses the Kriging interpolation method, which fully considers the correlation of spatial data, that is, the closer the monitoring points are, the stronger the deformation correlation. By constructing a semi-variogram model (such as a spherical model or an exponential model), spatial correlation is quantified, thereby realizing the deformation estimation of unobserved surface elements, ensuring that the spatial resolution of the interpolated BDS surface data is consistent with that of the GB-InSAR data, and maintaining the continuity of spatial deformation.

[0028] Step S203: Adjust the sampling rate of the planar grid data and the global planar deformation time series data to be consistent through linear interpolation, and synchronize the time.

[0029] In this embodiment, the GB-InSAR data sampling rate is ≥10Hz, and the acquired global planar deformation time series data belongs to high-frequency data. The BDS data sampling rate is usually configured to 10Hz. To achieve synchronization of the two types of data in the time dimension, a linear interpolation method is used to increase the sampling rate of the BDS planar grid data to the same frequency as the GB-InSAR global planar deformation time series data.

[0030] The specific interpolation principle is based on the assumption that the deformation between two adjacent BDS sampling points changes linearly, using the formula... The deformation value at the interpolation time is calculated, where t1 and t2 are adjacent sampling times, x1 and x2 are the corresponding deformation values, and t is the interpolation time. This process ensures that the sampling density of the two types of data is consistent in the time dimension, achieving timestamp alignment.

[0031] Optionally, after time synchronization is completed, the data from manual monitoring at the bridge's fixed supports is used as a benchmark to correct for deviations in the spatiotemporally registered data. Specifically, this involves extracting the deformation sequence at the corresponding location of the fixed supports from the spatiotemporally registered data, calculating its average value as a systematic deviation value, and then performing an overall translation correction on all registered data based on this systematic deviation value. It must be ensured that the deviation value is less than or equal to 0.3 mm. If it exceeds this threshold, the cause must be investigated and the calculation recalculated to eliminate systematic errors.

[0032] The Matlab function `interp1` can be used to perform linear interpolation on low-sampling-rate areal grid data, increasing its sampling rate to match that of GB-InSAR high-frequency data, thus achieving time synchronization between the two types of data. Spatially, the `kriging` interpolation function maps BDS discrete-point deformation data to GB-InSAR areal grid data. This interpolation method fully considers the correlation of spatial data, ensuring that the interpolated areal grid data has the same spatial resolution as the GB-InSAR data and maintains the continuity of spatial deformation. Through this dual temporal and spatial interpolation process, spatiotemporal registration of BDS and GB-InSAR data is achieved, laying the foundation for subsequent multi-source data fusion.

[0033] Step S204: Perform multi-source fusion estimation on the planar grid data and the global planar deformation time series data to generate the fused deformation field.

[0034] It is worth noting that the core of data fusion can fully leverage the areal coverage advantage of GB-InSAR and the high-precision point monitoring advantage of BDS. This project uses the Kalman filter algorithm to achieve real-time, high-precision estimation of dynamic deformation, outputting a globally consistent fused deformation field. Kalman filtering is a recursive optimal estimation method based on the state equation and observation equation of a linear system. Through an iterative prediction-update process, it combines prior state estimates with current observation data to continuously correct the state estimates, ultimately obtaining the posterior optimal estimate. This algorithm is suitable for real-time monitoring of dynamic systems (such as bridge deformation and landslide movement), featuring high computational efficiency, strong noise resistance, and the ability to effectively fuse complementary information from multi-source observation data.

[0035] This embodiment constructs a state vector containing displacement, velocity, and acceleration components; vertically corrects the global planar deformation time series data to obtain the corresponding vertical deformation sequence, and uses the vertical deformation sequence and the vertical deformation components in the planar grid data together as the observation vector for Kalman filtering; based on the optimal state estimate and state transition matrix of the previous moment, the prior state estimate of the current moment is calculated, and the Kalman gain is calculated using the observation vector to update the prior state estimate, obtaining the posterior optimal state estimate of the current moment; the displacement component in the posterior optimal state estimate is extracted as the fused deformation field.

[0036] For example, in this embodiment, the process of fusing global planar deformation time-series data and planar mesh data using Kalman filtering includes: Construct a three-dimensional state vector X containing displacement, velocity, and acceleration. k =[d k ,v k ,a k ]T, design the state equation Xk and observation equation Z k :

[0037] in, Let k represent the system state vector at time k. Represents the state transition matrix. This represents the system state vector at time k-1. Indicates process noise. Denotes the observation vector at time k. Represents the observation matrix. This indicates the observation noise, which is set according to the sensor's accuracy.

[0038] The global areal deformation time series data is in the LOS direction (the oblique angle θ between the radar line of sight and the ground surface). It needs to be converted into vertical (perpendicular to the ground surface) deformation data, which is more relevant to engineering applications. Therefore, the global areal deformation time series data is vertically corrected to convert it into the vertical deformation sequence that is relevant to engineering applications.

[0039] The vertical deformation sequence obtained by vertical correction, together with the vertical deformation component sequence in the time-synchronized planar grid data, is input into the Kalman filter fusion observation equation. Kalman filter prediction-update iteration is performed to complete the forward filtering solution, and finally the fused deformation field is output.

[0040] In the prediction phase, the posterior optimal state estimate is based on the previous time step (k-1 time step). Using the state transition matrix, calculate the prior state estimate at the current time (time k). :

[0041] At the same time, update the corresponding prior estimate error covariance matrix. :

[0042] in, Table of process noise covariance matrix.

[0043] During the update phase, the Kalman gain is calculated. :

[0044] in, Represents the observation matrix. This represents the observation noise covariance matrix.

[0045] Combining observation vectors from both GB-InSAR and BDS sources at the current time By using Kalman gain to correct the prior state estimate, the posterior optimal state estimate at the current time (time k) is obtained. :

[0046] And update the posterior estimation error covariance matrix. :

[0047] Where I is the identity matrix.

[0048] The above prediction and update iterations are repeated for each computational surface element (corresponding to an independent state vector) across the entire bridge domain. Finally, the displacement components in the posterior optimal state estimate at each time and location are extracted, resulting in the global, fused, high-precision vertical deformation time series data, which is the final global fused deformation field.

[0049] In this embodiment, the root mean square error (RMSE) can be used to verify the accuracy of the results. RMSE is used to measure the average deviation between the fused deformation field and the reference true value. The smaller the value, the higher the degree of agreement between the fused result and the real deformation field, that is, the better the fusion quality.

[0050] Based on the aforementioned steps, the state equations and observation equations required for Kalman filtering are constructed. In practical calculations, the `kalman` function in MATLAB can be used to implement the prediction-update iterative loop of Kalman filtering. In the prediction phase, the algorithm predicts the prior state at the current time based on the fused state (displacement, velocity, acceleration) from the previous time step. In the update phase, the algorithm fuses the current time step's high-precision point deformation observations from BDS and area deformation observations from GB-InSAR, corrects the prior state, and finally outputs the optimal state estimate for the current time step, thereby generating a global fused deformation field. After the fusion process is completed, the `rmse` function (or the equivalent root mean square error) calculates the RMSE value between the fused deformation field and the reserved, higher-precision reference deformation field (or the known true value). By comparing and analyzing the magnitude of this RMSE value, an objective and quantitative assessment of the overall quality of this data fusion can be made.

[0051] Step S3: Based on the fused deformation field, calculate multiple damage characteristic indicators that reflect different deformation physical mechanisms of the bridge, and determine the corresponding preliminary damage level based on the damage judgment threshold of each damage characteristic indicator.

[0052] In this embodiment, the damage characteristic indicators include the fused coordinate modal confidence criterion F-COMAC, the fused intra-integral volume change rate F-IAV, and the deformation gradient anomaly value DGA.

[0053] The calculation process of the fused coordinate modal confidence criterion F-COMAC includes, based on the mode shapes of the bridge in its current state and in a healthy baseline state, and combined with the strain energy weights that increase with the mode order, calculating the fused coordinate modal confidence criterion F-COMAC for each node of the bridge model:

[0054] In the formula, n is the total number of modal orders of the bridge structure, i is the index variable of the modal order, and j is the number of the bridge structure node. The strain energy weight corresponding to the i-th mode is defined as follows: =i / n is a weighting strategy that increases with the modal order. Its purpose is to give higher-order modes, which are more sensitive to changes in local stiffness, a higher weight in the index calculation, thereby enhancing the ability to identify damage. This represents the mode amplitude of the i-th mode at the j-th node under healthy baseline conditions of the bridge structure. This represents the mode amplitude of the i-th mode at the j-th node in the current (monitored) state of the bridge structure. The calculation results are located in the interval [0,1]. The closer the value is to 0, the more significant the change in the modal shape at node j is relative to the healthy state, and the greater the likelihood of damage to the node. Conversely, the closer the value is to 1, the better the modal characteristics of the node are, and the lower the likelihood of damage.

[0055] The calculation process of the fusion internal accumulation rate includes: extracting the deflection sequence reflecting the structural dynamic response from the fusion deformation field; performing layered processing on the deflection sequence; and extracting the first four high-frequency components. These four high-frequency components contain the main dynamic response information of the structure under environmental excitation and are highly sensitive to stiffness changes caused by damage. The current fusion internal accumulation is calculated based on the high-frequency components, which characterizes the total energy or intensity of the high-frequency dynamic response within the current time period. Based on the current fusion internal accumulation and the fusion internal accumulation under the bridge's healthy baseline state, the fusion internal accumulation rate is calculated.

[0056] in, This indicates the current accumulated amount within the fusion. This represents the cumulative amount of fusion within the bridge under the baseline health condition.

[0057] The Fusion Intra-Intrinsic Energy Variation Rate (F-IAV) is a rate of change index that reflects the relative change in the accumulated energy of the current dynamic response compared to a healthy baseline state. The larger the absolute value of F-IAV, the more significant the change in the dynamic response characteristics of that region, and the higher the probability of damage.

[0058] In this embodiment, the calculation of the Deformation Gradient Anomaly (DGA) aims to identify regions of sudden changes in local stiffness by analyzing the spatial rate of change of the fused deformation field. The calculation process is as follows: Taking advantage of the spatial continuity of GB-InSAR areal data, spatial gradient calculations are performed on the fused deformation field. This calculation can be performed using the formula DGA = The (d_F) implementation physically calculates the derivative (rate of change) of the deformation field along the spatial direction at each location, thus obtaining the global deformation gradient distribution field. In practical numerical calculations, this can be efficiently accomplished using MATLAB's gradient function or other equivalent gradient calculation tools.

[0059] When a bridge structure is in a healthy state, the deformation gradient field typically exhibits a gentle, continuous spatial distribution. However, when localized damage occurs (such as cracks or stiffness degradation), the stiffness of the damaged area abruptly changes, causing its deformation to become inconsistent with the surrounding area. This manifests as localized peaks or abnormally high value regions in the deformation gradient distribution field. These abnormal peaks are important spatial features for identifying potential damage.

[0060] This embodiment achieves gradient anomaly detection by setting a deformation gradient anomaly threshold. First, the mean μ and standard deviation σ of the deformation gradient distribution of the bridge under healthy baseline conditions are calculated. Then, the anomaly threshold is set to μ + 3σ (i.e., the healthy gradient mean plus three times the standard deviation). The deformation gradient distribution calculated under the current state is compared point-by-point with the aforementioned deformation gradient anomaly threshold. Regions with gradient values ​​exceeding the deformation gradient anomaly threshold (μ + 3σ) are identified as deformation gradient anomalies (DGA). Through this process, preliminary spatial localization of the damaged area is achieved.

[0061] Furthermore, in step S3, based on the damage judgment thresholds of each damage characteristic index, the corresponding preliminary damage level is determined, specifically including: The fused coordinate modal confidence criteria are compared with the corresponding confidence thresholds at each level to obtain the corresponding damage level. The confidence thresholds at each level are set based on the mean and standard deviation of the fused coordinate modal confidence criteria under the bridge health baseline state.

[0062] For example, the mean μ1 and standard deviation σ1 of the fused coordinate modal confidence criterion F-COMAC values ​​of all structural nodes under the bridge health baseline state are calculated. Subsequently, the method of subtracting a multiple of the standard deviation from the mean is used to set three levels of damage judgment thresholds: mild damage threshold: T1 = μ1 - 1.5σ1; moderate damage threshold: T2 = μ1 - 2.5σ1; severe damage threshold: T3 = μ1 - 3.5σ1.

[0063] The F-COMAC values ​​of each node's fused coordinate modal confidence criterion calculated under the current state are compared with the above three threshold levels to determine their initial damage level: if F-COMAC > T1, the corresponding initial damage level is determined to be intact; if T2 < F-COMAC ≤ T1, the corresponding initial damage level is determined to be mild; if T3 < F-COMAC ≤ T2, the corresponding initial damage level is determined to be moderate; if F-COMAC ≤ T3, the corresponding initial damage level is determined to be severe.

[0064] The F-COMAC value ranges from [0, 1]. The closer the value is to 0, the greater the difference between the mode shape of the node and the healthy baseline state, and the higher the probability of damage. The above threshold determination rule transforms this continuous probability index into a discrete damage level, realizing the transformation from index quantification to preliminary classification of damage severity.

[0065] Simultaneously, the rate of change of the accumulated volume within the fusion is compared with the corresponding rate of change thresholds at each level to obtain the corresponding damage level, wherein the rate of change thresholds at each level are set based on the rate of change of the accumulated volume within the fusion under the bridge health baseline state.

[0066] For example, the F-IAV (Functional Energy Change Rate) index, based on a multiple of the maximum absolute value of the bridge's health baseline state, is used to calibrate the classification threshold. F-IAV is a rate-of-change index; the larger its absolute value, the more drastic the change in the accumulated energy of the current dynamic response relative to the healthy state, and the more severe the corresponding regional damage. First, the maximum absolute value of F-IAV at all monitoring points is calculated for the bridge under the healthy baseline state, denoted as IAV_max. Then, based on this maximum value, the following three damage judgment thresholds are set: Mild damage threshold: T4 = 1.2 × IAV_max; Moderate damage threshold: T5 = 1.5 × IAV_max; Severe damage threshold: T6 = 2.0 × IAV_max.

[0067] The absolute value of the fusion accumulation change rate F-IAV at each location, calculated under the current monitoring status, |F-IAV|, is compared with the three threshold levels mentioned above to determine the initial damage level: if |F-IAV|≤T4, the corresponding initial damage level is determined to be intact; if T4<|F-IAV|≤T5, the corresponding initial damage level is determined to be mild; if T5<|F-IAV|≤T6, the corresponding initial damage level is determined to be moderate; if |F-IAV|>T6, the corresponding initial damage level is determined to be severe.

[0068] The F-IAV (Fault-Induced Impact) determination rule is based on its absolute value, reflecting the direction-insensitive nature of the cumulative energy change in the dynamic response caused by damage, while the magnitude of the change is key to measuring the severity of the damage. The aforementioned threshold determination rule transforms the continuous numerical index reflecting the relative change in energy accumulation into a discrete damage severity level, providing independent evidence input based on the F-IAV index for subsequent multi-index fusion decision-making.

[0069] Meanwhile, the abnormal values ​​of the deformation gradient are compared with the corresponding deformation gradient thresholds at each level to obtain the corresponding damage level, wherein the deformation gradient thresholds at each level are set based on the mean and standard deviation of the deformation gradient under the bridge's healthy baseline state.

[0070] For example, the deformation gradient outlier (DGA) index is calibrated using the 3σ statistical principle based on the statistical characteristics of the health status to determine the classification threshold. First, the mean μ² and standard deviation σ² of the deformation gradient distribution field (DGA field) of the bridge under a healthy baseline state are calculated. Then, based on this mean and standard deviation, the following four damage judgment thresholds are set: Suspected damage threshold: T7 = μ² + 3σ²; Mild damage threshold: T8 = μ² + 4σ²; Moderate damage threshold: T9 = μ² + 5σ²; Severe damage threshold: T10 = μ² + 6σ².

[0071] The deformation gradient anomaly value (DGA) calculated at each location under the current monitoring state is compared with the above four thresholds to determine its initial damage level: if DGA≤T7, the corresponding initial damage level is determined to be intact; if T7<DGA≤T8, the corresponding initial damage level is determined to be mild; if T8<DGA≤T9, the corresponding initial damage level is determined to be moderate; if DGA>T10, the corresponding initial damage level is determined to be severe.

[0072] The DGA index reflects the spatial rate of change of the structural deformation field; abrupt changes in local stiffness can lead to an abnormal increase in gradient values. The aforementioned rules achieve a quantitative mapping from abnormal gradient values ​​to damage levels, providing independent evidence input based on the DGA index for subsequent multi-index fusion decision-making.

[0073] Step S4: Use the preliminary damage levels of each damage characteristic index as independent sources of evidence, perform evidence fusion to resolve index conflicts, and determine the bridge damage location and damage level based on the fusion results.

[0074] In one embodiment, such as Figure 3 As shown, step S4 specifically includes the following steps: Step S401: Establish an identification framework containing multiple mutually exclusive damage levels. The identification framework includes four damage levels: intact, slightly damaged, moderately damaged, and severely damaged, covering all damage assessment scenarios. The identification framework is defined as the set Θ = {H0 (intact), H1 (slightly damaged), H2 (moderately damaged), H3 (severely damaged)}.

[0075] Step S402: The fused coordinate mode confidence criterion, the rate of change of fused internal accumulation, and the damage level corresponding to the deformation gradient anomaly value are respectively used as three independent sources of evidence.

[0076] Step S403: Construct a basic probability assignment function for each source of evidence. The basic probability assignment function assigns a corresponding basic probability value to each damage level according to the corresponding preliminary damage level, and sets a corresponding uncertainty probability for each source of evidence.

[0077] Specifically, F-COMAC, F-IAV, and DGA are treated as three independent sources of evidence, and a basic probability assignment function m1 is constructed accordingly. m2 ( ), m3 ( To quantify the inherent uncertainty of each evidence source due to monitoring noise, model error, and environmental interference, a fixed uncertainty coefficient αi (i=1,2,3) is assigned to each evidence source, and mi(Θ)=αi is directly set. This coefficient reflects the prior assessment of the reliability of the evidence source: F-COMAC (i=1): modal correlation index, with good stability and low noise, with an uncertainty coefficient α1∈[0.05,0.1]; F-IAV (i=2): vibration response index, with moderate noise, with an uncertainty coefficient α2∈[0.1,0.15]; DGA (i=3): dynamic gradient index, which is susceptible to environmental interference, with an uncertainty coefficient α3∈[0.15,0.2].

[0078] For the remaining probability mass (1 αi), assigned based on the preliminary damage assessment results of each evidence source. Specifically, a linear membership function is used to map the index value (or its assessment result) of each evidence source to a membership degree for the four single-level propositions H0, H1, H2, and H3 in the identification framework (corresponding to intact, mild, moderate, and severe damage, respectively). Subsequently, these membership degrees are normalized to obtain a set of normalized weights. Finally, (1 The probability quality of αi is allocated to the corresponding single-level propositions according to the proportion of this normalized weight.

[0079] Through the above allocation, a complete basic probability assignment function mi is constructed for each source of evidence. This function strictly satisfies ∑A Θmi(A)=1, where A is any subset of the recognition frame Θ, ensuring consistency in probability allocation while satisfying mi( )=0 means that the probability of an impossible event is zero.

[0080] Step S404: The basic probability assignment functions of the three evidence sources are fused and calculated using the evidence fusion rule to obtain the joint basic probability assignment.

[0081] Specifically, this embodiment employs the Dempster orthogonal synthesis rule in DS evidence synthesis to synthesize the basic probability assignment functions of two evidence sources, obtaining an intermediate joint basic probability assignment. This intermediate joint basic probability assignment is then combined with the basic probability assignment function of the third evidence source using Dempster synthesis again to obtain the final joint basic probability assignment. This achieves automatic resolution of indicator conflicts, correction of misjudgments based on single indicators, and realization of uncertainty fusion of multi-source information.

[0082] For example, the DS evidence synthesis operator satisfies the associative law. For the synthesis of three evidence sources, this embodiment adopts a two-step sequential fusion strategy to reduce the computational complexity of high-dimensional operations and ensure numerical stability: The first step is to perform Dempster orthogonal synthesis on the first two evidence sources m1 and m2 to calculate the intermediate joint basic probability assignment m. 12 =m1⊕m2; The second step is to assign m values ​​to the aforementioned inter-joint basic probabilities. 12 Performing a Dempster orthogonal synthesis with the third source of evidence m3 again yields the final joint fundamental probability assignment BPA: m = m 12 ⊕m3.

[0083] For any two sources of evidence m1 and m2, the combined result m = m1 ⊕ m2 is defined by the following mathematical form: Calculate the conflict coefficient K to quantify the degree of conflict between two sources of evidence:

[0084] Here, K represents the sum of the probabilities of all pairs of subsets whose intersection is empty. The closer K is to 1, the stronger the conflict, indicating that the two sources of evidence support completely different propositions; K=0 indicates that the two sources of evidence are completely compatible and there is no conflict.

[0085] Subsequently, the joint BPA m(C) is calculated (where C... Θ is any non-empty subset):

[0086] In the formula, The conflict normalization coefficient is used to redistribute probability mass when there is conflicting evidence, eliminating contradictions caused by the conflict while strictly ensuring that the sum of the probability distributions after synthesis is 1; the summation term ∑m1(A) m2(B) iterates through all subset pairs (A,B) in the identification framework Θ that satisfy A∩B=C, multiplies and sums their probabilities to obtain the joint evidence strength that supports the intersection C.

[0087] Through the above synthesis mechanism, this method can automatically resolve potential conflicts between multiple indicators, reduce the risk of misjudgment caused by environmental interference or local ambiguity of a single indicator, and ultimately achieve high-confidence comprehensive damage judgment based on multi-source heterogeneous monitoring data.

[0088] Step S405: Select the damage level with the highest confidence as the final damage level for the corresponding position based on the joint basic probability assignment.

[0089] In the final joint basic probability assignment m, the confidence (probability value) of the four single-element damage level propositions is extracted respectively: m(H0), confidence for intact level; m(H1), confidence for mild damage level; m(H2), confidence for moderate damage level; m(H3), confidence for severe damage level.

[0090] For each computational unit of the bridge, the confidence scores for four single damage levels—intact, slightly damaged, moderately damaged, and severely damaged—are extracted from the basic probability assignment. The highest confidence score is selected as the final damage level for that computational unit. This confidence comparison and level determination process is independently repeated for all computational units across the entire bridge area. This results in a final damage level distribution map covering the entire bridge area, clearly indicating the specific location of the damage and its corresponding damage level.

[0091] In a preferred implementation, the maximum confidence level of the joint basic probability assignment is compared with a preset confidence threshold. If the maximum confidence level is greater than or equal to the confidence threshold, the current evidence fusion result is considered to have sufficient reliability, and the damage level corresponding to the maximum confidence level is directly determined as the final damage level for that location. If the maximum confidence level is less than the confidence threshold, it is considered that there is a serious conflict between the current multi-source evidence, or that the monitoring noise interference is too large, resulting in insufficient confidence in the fusion decision. The decision result for that location is determined to be unreliable monitoring noise and is removed. In the final damage distribution map, this location may not output a specific level or may be marked as uncertain, thereby ensuring that every published damage determination has a high confidence level.

[0092] This application presents a method for precise identification and localization of bridge damage based on multi-source data fusion. By fusing complementary monitoring data from GB-InSAR and BDS, and utilizing DS evidence theory to intelligently fuse multi-dimensional damage indicators, it ultimately achieves high-precision, high-confidence spatial localization and grade determination of bridge damage. This method effectively overcomes the limitations of single-sensor monitoring, resolves the conflict problem in multi-indicator judgment, and significantly improves the accuracy and reliability of damage identification.

[0093] The obtained damage identification results provide direct and reliable input for the digital and refined operation and maintenance of bridge structures. Based on this high-confidence damage status diagnosis, the bridge digital twin model can be further calibrated online, thereby supporting quantitative assessment of structural safety and prediction of remaining life. This achieves a complete technical closed loop from monitoring to diagnosis to prediction to maintenance, significantly improving the initiative, safety, and economy of bridge operation and maintenance management.

[0094] Secondly, embodiments of this application also provide a bridge damage precision identification and positioning device based on multi-source data fusion.

[0095] In one embodiment, reference is made to Figure 4 , Figure 4 This is a functional module diagram of an embodiment of the bridge damage precision identification and positioning device based on multi-source data fusion according to this application. Figure 4 As shown, the bridge damage precision identification and location device based on multi-source data fusion includes: The acquisition module is used to acquire the time-series data of the bridge's global planar deformation through ground-based synthetic aperture radar and the time-series data of the bridge's point-like three-dimensional deformation through the BeiDou satellite navigation system. The fusion module is used to perform spatiotemporal registration and data fusion on the global planar deformation time series data and the point-like three-dimensional deformation time series data to obtain the fused deformation field of the entire bridge area. The calculation module is used to calculate multiple damage characteristic indicators reflecting different deformation physical mechanisms of the bridge based on the fused deformation field, and to determine the corresponding preliminary damage level based on the damage judgment threshold of each damage characteristic indicator. The damage characteristic indicators include the fused coordinate mode confidence criterion, the rate of change of fused internal accumulation, and the deformation gradient anomaly value. The determination module is used to use the preliminary damage levels of each damage characteristic index as independent evidence sources, perform evidence fusion to resolve index conflicts, and determine the location and level of bridge damage based on the fusion results.

[0096] Furthermore, in one embodiment, the fusion module is also used for: The global planar deformation time series data and the point-like three-dimensional deformation time series data are unified into the geodetic coordinate system; The point-like three-dimensional deformation time-series data is mapped to the monitoring surface element of the ground-based synthetic aperture radar by spatial interpolation, and the point-like three-dimensional deformation time-series data is converted into surface grid data with the same spatial resolution as the global surface deformation time-series data. The sampling rates of the planar grid data and the global planar deformation time series data are adjusted to be consistent through linear interpolation, and time synchronization is performed. The planar grid data and the global planar deformation time series data are subjected to multi-source fusion estimation to generate the fused deformation field.

[0097] Furthermore, in one embodiment, the fusion module is also used for: Based on the spatial distribution characteristics of the point-like three-dimensional deformation time series data, the experimental semivariogram is calculated and fitted by a preset theoretical model to obtain the theoretical semivariogram model. For each monitoring surface element to be mapped, a set of Kriging equations is constructed based on the theoretical semi-variogram model, and the interpolation weights of the monitoring points of each point-like three-dimensional deformation time series data relative to the monitoring surface element are solved. The point-like three-dimensional deformation time series data are weighted and summed according to the interpolation weights to calculate the deformation estimate of the corresponding monitoring surface element; Traverse all monitored surface elements to generate surface mesh data with the same spatial resolution as the global surface deformation time series data.

[0098] Furthermore, in one embodiment, the fusion module is also used for: Construct a state vector containing displacement, velocity, and acceleration components; The global planar deformation time series data is vertically corrected to obtain the corresponding vertical deformation sequence, and the vertical deformation sequence and the vertical deformation component sequence in the planar grid data are used together as the observation vector of the Kalman filter. Based on the optimal state estimate and state transition matrix of the previous time step, the prior state estimate of the current time step is calculated, and the Kalman gain is calculated using the observation vector to update the prior state estimate, so as to obtain the posterior optimal state estimate of the current time step. The displacement components in the posterior optimal state estimate are extracted as the fused deformation field.

[0099] Furthermore, in one embodiment, the computing module is also used for: Based on the mode shapes of the bridge under its current and healthy baseline conditions, and combined with the strain energy weights that increase with the mode order, the fused coordinate mode confidence criteria for each node of the bridge model are calculated:

[0100] in, The fused coordinate modal confidence criterion is defined as follows: n is the total number of modal orders of the bridge structure, i is the index variable of the modal order, and j is the node number of the bridge structure. The strain energy weights are the values ​​corresponding to the i-th mode. This represents the mode amplitude of the i-th mode at the j-th node under healthy baseline conditions of the bridge structure. This represents the mode amplitude of the i-th mode at the j-th node in the current state of the bridge structure; The deflection sequence in the fused deformation field is processed in layers to extract the high-frequency components. The current fused internal accumulation is calculated based on the high-frequency components. The rate of change of the fused internal accumulation is calculated based on the current fused internal accumulation and the fused internal accumulation under the bridge health baseline state. Spatial gradient calculation is performed on the fused deformation field to obtain the deformation gradient distribution. Based on the deformation gradient distribution and a preset deformation gradient anomaly threshold, the deformation gradient anomaly value is determined.

[0101] Furthermore, in one embodiment, the computing module is also used for: The fused coordinate mode confidence criterion is compared with the corresponding confidence thresholds at each level to obtain the corresponding preliminary damage level. The confidence thresholds at each level are set based on the mean and standard deviation of the fused coordinate mode confidence criterion under the bridge health baseline state. The change rate of the accumulated volume within the fusion is compared with the corresponding change rate thresholds at each level to obtain the corresponding preliminary damage level, wherein the change rate thresholds at each level are set based on the change rate of the accumulated volume within the fusion under the bridge health baseline state. The abnormal values ​​of the deformation gradient are compared with the corresponding deformation gradient thresholds at each level to obtain the corresponding preliminary damage level, wherein the deformation gradient thresholds at each level are set based on the mean and standard deviation of the deformation gradient under the bridge health baseline state. The preliminary damage level includes intact, minor damage, moderate damage, or severe damage.

[0102] Furthermore, in one embodiment, the determining module is further configured to: A recognition framework is established that includes multiple mutually exclusive damage levels, comprising four damage levels: intact, slightly damaged, moderately damaged, and severely damaged. The fused coordinate mode confidence criterion, the rate of change of fused internal accumulation, and the preliminary damage level corresponding to the deformation gradient anomaly value are respectively used as three independent sources of evidence; A basic probability assignment function is constructed for each source of evidence. Based on the corresponding preliminary damage level, the basic probability assignment function assigns a corresponding basic probability value to each damage level and sets a corresponding uncertainty probability for each source of evidence. The basic probability assignment functions of the three evidence sources are fused and calculated using the evidence fusion rule to obtain the joint basic probability assignment. The damage level with the highest confidence level is selected as the final damage level at the corresponding location based on the joint basic probability assignment.

[0103] Furthermore, in one embodiment, the determining module is further configured to: Based on the orthogonal synthesis rule, the basic probability assignment functions of two evidence sources are synthesized to obtain an intermediate joint basic probability assignment. The intermediate joint basic probability assignment and the basic probability assignment function of the third source of evidence are orthogonally combined again to obtain the final joint basic probability assignment.

[0104] Furthermore, in one embodiment, the determining module is further configured to: Compare the maximum confidence level in the joint basic probability assignment with the preset confidence threshold; If the maximum confidence level is greater than or equal to the confidence threshold, then the damage level corresponding to the maximum confidence level is determined as the final damage level. If the maximum confidence level is less than the confidence level threshold, the result is determined to be monitoring noise and is removed.

[0105] The functions of each module in the bridge damage accurate identification and positioning device based on multi-source data fusion correspond to the steps in the embodiment of the bridge damage accurate identification and positioning method based on multi-source data fusion. Their functions and implementation processes will not be described in detail here.

[0106] It should be noted that the sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0107] The terms "comprising" and "having," and any variations thereof, in the specification, claims, and accompanying drawings of this application are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to such process, method, product, or apparatus. The terms "first," "second," and "third," etc., are used to distinguish different objects, etc., and do not indicate a sequence, nor do they limit "first," "second," and "third" to different types.

[0108] In the description of the embodiments of this application, terms such as "exemplary," "for example," or "for instance" are used to indicate examples, illustrations, or explanations. Any embodiment or design described as "exemplary," "for example," or "for instance" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of terms such as "exemplary," "for example," or "for instance" is intended to present the relevant concepts in a concrete manner.

[0109] In the description of the embodiments of this application, unless otherwise stated, " / " means "or". For example, A / B can mean A or B. The "and / or" in the text is merely a description of the relationship between related objects, indicating that there can be three relationships. For example, A and / or B can mean: A exists alone, A and B exist simultaneously, and B exists alone. In addition, in the description of the embodiments of this application, "multiple" means two or more.

[0110] In some processes described in the embodiments of this application, multiple operations or steps are included in a specific order. However, it should be understood that these operations or steps may not be executed in the order they appear in the embodiments of this application, or they may be executed in parallel. The sequence number of the operation is only used to distinguish different operations, and the sequence number itself does not represent any execution order. In addition, these processes may include more or fewer operations, and these operations or steps may be executed sequentially or in parallel, and these operations or steps may be combined.

[0111] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes several instructions to cause a terminal device to execute the methods described in the various embodiments of this application.

[0112] The above are merely preferred embodiments of this application and do not limit the patent scope of this application. Any equivalent structural or procedural transformations made using the content of this application's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.

Claims

1. A method for accurate identification and localization of bridge damage based on multi-source data fusion, characterized in that, The method includes: The bridge's global planar deformation time-series data were obtained by ground-based synthetic aperture radar, and the bridge's point-like three-dimensional deformation time-series data were obtained by the BeiDou satellite navigation system. The spatiotemporal registration and data fusion of the global planar deformation time series data and the point-like three-dimensional deformation time series data are performed to obtain the fused deformation field of the entire bridge. Based on the fused deformation field, multiple damage characteristic indicators reflecting different deformation physical mechanisms of the bridge are calculated, and based on the damage judgment threshold of each damage characteristic indicator, the corresponding preliminary damage level is determined. The damage characteristic indicators include the fused coordinate mode confidence criterion, the rate of change of fused internal accumulation, and the deformation gradient anomaly value. The preliminary damage levels of each damage characteristic index are used as independent sources of evidence. Evidence fusion is performed to resolve index conflicts, and the location and final damage level of the bridge damage are determined based on the fusion results.

2. The bridge damage accurate identification and location method based on multi-source data fusion as described in claim 1, characterized in that, The process of spatiotemporal registration and data fusion of global planar deformation time-series data and point-based three-dimensional deformation time-series data to obtain the fused deformation field of the entire bridge includes: The global planar deformation time series data and the point-like three-dimensional deformation time series data are unified into the geodetic coordinate system; The point-like three-dimensional deformation time-series data is mapped to the monitoring surface element of the ground-based synthetic aperture radar by spatial interpolation, and the point-like three-dimensional deformation time-series data is converted into surface grid data with the same spatial resolution as the global surface deformation time-series data. The sampling rates of the planar grid data and the global planar deformation time series data are adjusted to be consistent through linear interpolation, and time synchronization is performed. The planar grid data and the global planar deformation time series data are subjected to multi-source fusion estimation to generate the fused deformation field.

3. The bridge damage accurate identification and location method based on multi-source data fusion as described in claim 2, characterized in that, The point-like three-dimensional deformation time-series data is mapped to the monitoring elements of the ground-based synthetic aperture radar through spatial interpolation, and the point-like three-dimensional deformation time-series data is converted into planar mesh data with the same spatial resolution as the global planar deformation time-series data, including: Based on the spatial distribution characteristics of the point-like three-dimensional deformation time series data, the experimental semivariogram is calculated and fitted by a preset theoretical model to obtain the theoretical semivariogram model. For each monitoring surface element to be mapped, a set of Kriging equations is constructed based on the theoretical semi-variogram model, and the interpolation weights of the monitoring points of each point-like three-dimensional deformation time series data relative to the monitoring surface element are solved. The point-like three-dimensional deformation time series data are weighted and summed according to the interpolation weights to calculate the deformation estimate of the corresponding monitoring surface element; Traverse all monitored surface elements to generate surface mesh data with the same spatial resolution as the global surface deformation time series data.

4. The bridge damage accurate identification and location method based on multi-source data fusion as described in claim 2, characterized in that, The process of multi-source fusion estimation of the planar grid data and the global planar deformation time series data to generate the fused deformation field includes: Construct a state vector containing displacement, velocity, and acceleration components; The global planar deformation time series data is vertically corrected to obtain the corresponding vertical deformation sequence, and the vertical deformation sequence and the vertical deformation component sequence in the planar grid data are used together as the observation vector of the Kalman filter. Based on the optimal state estimate and state transition matrix of the previous time step, the prior state estimate of the current time step is calculated, and the Kalman gain is calculated using the observation vector to update the prior state estimate, so as to obtain the posterior optimal state estimate of the current time step. The displacement components in the posterior optimal state estimate are extracted as the fused deformation field.

5. The bridge damage accurate identification and location method based on multi-source data fusion as described in claim 1, characterized in that, Based on the fused deformation field, multiple damage characteristic indicators reflecting different deformation physical mechanisms of the bridge are calculated, including: Based on the mode shapes of the bridge under its current and healthy baseline conditions, and combined with the strain energy weights that increase with the mode order, the fused coordinate modal confidence criteria for each node of the bridge model are calculated: in, The fused coordinate modal confidence criterion is defined as follows: n is the total number of modal orders of the bridge structure, i is the index variable of the modal order, and j is the node number of the bridge structure. The strain energy weights are the values ​​corresponding to the i-th mode. This represents the mode amplitude of the i-th mode at the j-th node under healthy baseline conditions of the bridge structure. This represents the mode amplitude of the i-th mode at the j-th node in the current state of the bridge structure; The deflection sequence in the fused deformation field is processed in layers to extract the high-frequency components. The current fused internal accumulation is calculated based on the high-frequency components. The rate of change of the fused internal accumulation is calculated based on the current fused internal accumulation and the fused internal accumulation under the bridge health baseline state. Spatial gradient calculation is performed on the fused deformation field to obtain the deformation gradient distribution. Based on the deformation gradient distribution and the preset deformation gradient anomaly threshold, the deformation gradient anomaly value is determined.

6. The bridge damage accurate identification and location method based on multi-source data fusion as described in claim 5, characterized in that, The determination of the corresponding preliminary damage level based on the damage judgment thresholds of various damage characteristic indicators includes: The fused coordinate mode confidence criterion is compared with the corresponding confidence thresholds at each level to obtain the corresponding preliminary damage level. The confidence thresholds at each level are set based on the mean and standard deviation of the fused coordinate mode confidence criterion under the bridge health baseline state. The change rate of the accumulated volume within the fusion is compared with the corresponding change rate thresholds at each level to obtain the corresponding preliminary damage level, wherein the change rate thresholds at each level are set based on the change rate of the accumulated volume within the fusion under the bridge health baseline state. The abnormal values ​​of the deformation gradient are compared with the corresponding deformation gradient thresholds at each level to obtain the corresponding preliminary damage level, wherein the deformation gradient thresholds at each level are set based on the mean and standard deviation of the deformation gradient under the bridge health baseline state. The preliminary damage level includes intact, minor damage, moderate damage, or severe damage.

7. The bridge damage accurate identification and location method based on multi-source data fusion as described in claim 6, characterized in that, The process of using the preliminary damage levels of each damage characteristic index as independent sources of evidence, fusing evidence to resolve index conflicts, and determining the bridge damage location and damage level based on the fusion results includes: A recognition framework is established that includes multiple mutually exclusive damage levels, comprising four damage levels: intact, slightly damaged, moderately damaged, and severely damaged. The fused coordinate mode confidence criterion, the rate of change of fused internal accumulation, and the preliminary damage level corresponding to the deformation gradient anomaly value are respectively used as three independent sources of evidence; A basic probability assignment function is constructed for each source of evidence. Based on the corresponding preliminary damage level, the basic probability assignment function assigns a corresponding basic probability value to each damage level and sets a corresponding uncertainty probability for each source of evidence. The basic probability assignment functions of the three evidence sources are fused and calculated using the evidence fusion rule to obtain the joint basic probability assignment. The damage level with the highest confidence level is selected as the final damage level at the corresponding location based on the joint basic probability assignment.

8. The bridge damage accurate identification and location method based on multi-source data fusion as described in claim 7, characterized in that, The process of fusing and calculating the basic probability assignment functions of the three evidence sources using evidence fusion rules to obtain the joint basic probability assignment includes: Based on the orthogonal synthesis rule, the basic probability assignment functions of two evidence sources are synthesized to obtain an intermediate joint basic probability assignment. The intermediate joint basic probability assignment and the basic probability assignment function of the third source of evidence are orthogonally combined again to obtain the final joint basic probability assignment.

9. The bridge damage accurate identification and location method based on multi-source data fusion as described in claim 7, characterized in that, The damage level with the highest confidence level is selected as the final damage level at the corresponding location based on the joint basic probability assignment, including: Compare the maximum confidence level in the joint basic probability assignment with the preset confidence threshold; If the maximum confidence level is greater than or equal to the confidence threshold, then the damage level corresponding to the maximum confidence level is determined as the final damage level. If the maximum confidence level is less than the confidence level threshold, the result is determined to be monitoring noise and is removed.

10. A bridge damage precision identification and positioning device based on multi-source data fusion, characterized in that, The device includes: The acquisition module is used to acquire the time-series data of the bridge's global planar deformation through ground-based synthetic aperture radar and the time-series data of the bridge's point-like three-dimensional deformation through the BeiDou satellite navigation system. The fusion module is used to perform spatiotemporal registration and data fusion on the global planar deformation time series data and the point-like three-dimensional deformation time series data to obtain the fused deformation field of the entire bridge area. The calculation module is used to calculate multiple damage characteristic indicators reflecting different deformation physical mechanisms of the bridge based on the fused deformation field, and to determine the corresponding preliminary damage level based on the damage judgment threshold of each damage characteristic indicator. The damage characteristic indicators include the fused coordinate mode confidence criterion, the rate of change of fused internal accumulation, and the deformation gradient anomaly value. The determination module is used to use the preliminary damage levels of each damage characteristic index as independent evidence sources, perform evidence fusion to resolve index conflicts, and determine the location and level of bridge damage based on the fusion results.

Images