A bridge structure health analysis method and system based on vibration signals

Abstract

The application discloses a bridge structure health analysis method and system based on vibration signals, comprising: deploying a vibration sensor array at a plurality of key monitoring nodes of a bridge structure, and synchronizing output signals of the vibration sensor array to a central processing unit; selecting a preset proportion of key sensors from the vibration sensor array as master nodes, and assigning initial weights to the master nodes; based on the weights of the master nodes and other sensors, processing the collected vibration signals using an adaptive filtering algorithm, and generating a comprehensive feature vector according to the filtered vibration signals; according to the comprehensive feature vector, constructing a bridge structure damage distribution model to simulate the dynamic evolution process of damage in the bridge structure; and when the damage distribution model converges, mapping the comprehensive feature vector to a two-dimensional space to generate a bridge structure damage distribution map. The application can realize accurate identification, dynamic evolution simulation and intuitive visualization of bridge structure damage.

Description

Technical Field

[0001] This invention relates to the field of civil engineering structural health monitoring technology, and in particular to a method and system for bridge structural health analysis based on vibration signals. Background Technology

[0002] As a critical transportation infrastructure, the structural health of bridges is directly related to public safety. During long-term service, bridge structures are prone to cumulative damage under the combined effects of multiple factors such as vehicle loads, environmental corrosion, and material fatigue.

[0003] Traditional methods for assessing the health of bridge structures often employ offline testing, periodic manual inspections, or fixed-point static monitoring. Offline testing typically involves professionals using equipment to inspect key components of the bridge point-by-point while it is either out of service or partially out of service. This method is time-consuming, costly, and results in a significant delay, failing to capture changes in the bridge's structural condition under dynamic loads. Periodic manual inspections are conducted at fixed intervals, but are limited by the sample size and scope of inspection, making it difficult to comprehensively reflect the overall health of the bridge structure and prone to missing defects, especially early, minor damage in hidden areas. Static monitoring introduces automated data acquisition and simple feedback control, but these mechanisms are often based on preset rules and lack the ability to perceive and make intelligent decisions regarding the bridge's dynamic service status. For example, some vibration signal analysis methods based on fixed models use fixed parameters and algorithms after model training. When new types of damage occur in the bridge structure or the service environment changes, the model's analytical accuracy drops significantly. Moreover, existing feedback systems can usually only analyze a single bridge component or a single type of damage, making it difficult to achieve collaborative optimization of the detection of multiple types of damage across the entire bridge structure, and thus unable to assess the overall health status of the bridge structure. Summary of the Invention

[0004] The purpose of this invention is to provide a method and system for bridge structural health analysis based on vibration signals. This invention can achieve accurate identification of bridge structural damage, dynamic evolution simulation, and intuitive visualization.

[0005] The technical solution of this invention: A method for bridge structural health analysis based on vibration signals, comprising the following steps:

[0006] Step 1: Deploy vibration sensor arrays at multiple key monitoring nodes of the bridge structure and synchronize the output signals of the vibration sensor arrays to the central processing unit;

[0007] Step 2: Select a preset proportion of key sensors from the vibration sensor array as master control nodes and assign them initial weights;

[0008] Step 3: Based on the weights of the master control node and other sensors, the collected vibration signal is processed using an adaptive filtering algorithm, and a comprehensive feature vector is generated based on the filtered vibration signal.

[0009] Step 4: Based on the comprehensive feature vector, construct a bridge structure damage distribution model to simulate the dynamic evolution of damage in the bridge structure;

[0010] Step 5: After the damage distribution model converges, the comprehensive feature vector is mapped to a two-dimensional space to generate a bridge structure damage distribution map.

[0011] In the above-mentioned bridge structural health analysis method based on vibration signals, the core parameters in step 1 include at least the vibration signal sampling frequency, data resolution, feature extraction window length, feedback control cycle, environmental compensation coefficient, and damage identification threshold.

[0012] In the aforementioned bridge structural health analysis method based on vibration signals, step 3, the mathematical expression of the adaptive filtering algorithm is:

[0013] ;

[0014] In the formula: This is the filtered output signal. For time, For the first The weighting coefficients of each sensor; For the first The raw vibration signal collected by each sensor For bias terms, This represents the total number of sensors;

[0015] The weighting coefficient The update is performed using the minimum mean square error criterion, and the update formula is as follows:

[0016] ;

[0017] In the formula: For learning rate, This represents the error between the filtered output and the target value.

[0018] In the aforementioned bridge structural health analysis method based on vibration signals, step 3, the step of generating the comprehensive feature vector, specifically includes:

[0019] The contribution of each vibration sensor is calculated based on the initial weights, and the contribution is normalized to the range of 0 to 1.

[0020] Sensors with a normalized contribution greater than 0.5 are marked as high-priority nodes, and the remaining sensors are marked as low-priority nodes.

[0021] The vibration signals of high-priority nodes are processed by weighted averaging to generate enhanced feature vectors;

[0022] The vibration signals of low-priority nodes are denoised to remove environmental noise and traffic noise interference, and a denoised feature vector is generated.

[0023] The enhanced feature vector is merged with the denoised feature vector to generate the comprehensive feature vector.

[0024] In the aforementioned bridge structural health analysis method based on vibration signals, step 4 involves using the comprehensive feature vector to evaluate the convergence of the damage distribution model and performing iterative processing if convergence fails. The damage distribution model simulates the dynamic evolution of damage in the bridge structure, and its mathematical model is as follows:

[0025] ;

[0026] In the formula: Let be the damage density distribution function. For time, The damage diffusion coefficient is... For the Laplace operator, The source term represents the damage generation rate. The calculation relies on the comprehensive feature vector, which is used to transform the characteristics of the vibration signal into the driving force for damage generation.

[0027] The aforementioned bridge structural health analysis method based on vibration signals calculates the convergence index of the damage distribution model based on the comprehensive feature vector. The calculation formula is as follows:

[0028] ;

[0029] In the formula: As a convergence index, For the first Each feature vector in time The value, The total number of eigenvectors;

[0030] If the convergence index is less than the set threshold, the construction steps of the repeated damage distribution model with the current comprehensive feature vector as input will continue until the convergence index reaches or exceeds the set threshold.

[0031] The aforementioned bridge structural health analysis method based on vibration signals includes the following step: mapping feature vectors to a two-dimensional space to generate a bridge structural damage distribution map.

[0032] Damage index extraction: Based on the comprehensive feature vector, calculate the damage index corresponding to each sensor location;

[0033] Spatial coordinate correlation: Obtain the two-dimensional planar coordinates of each vibration sensor on the bridge structure;

[0034] Two-dimensional interpolation calculation: Based on the two-dimensional plane coordinates, the damage index is processed using a spatial interpolation algorithm to generate a continuous damage index distribution matrix covering the two-dimensional plane of the bridge;

[0035] Generating a heat map: The values ​​of the damage index distribution matrix are mapped to a preset color gradient to generate a heat map of the bridge structure damage distribution.

[0036] The aforementioned bridge structural health analysis method based on vibration signals uses a damage index that is a quantitative correlation value between vibration signal characteristics and the degree of bridge structural damage. This index is calculated based on a comprehensive feature vector, and the formula is as follows:

[0037] ;

[0038] In the formula, For the first The damage index at each sensor monitoring location; the higher the value, the greater the degree of structural damage at that location. For the first The combined feature vector of each sensor, No. The baseline integrated feature vector of the bridge under undamaged conditions at each sensor location. , Both are weighting coefficients, and their sum is 1; For the first The standard deviation of the combined feature vector of each sensor characterizes the dispersion of the vibration signal; the higher the dispersion, the worse the structural stability.

[0039] The aforementioned bridge structural health analysis method based on vibration signals is based on damage indicators. A continuous damage index distribution matrix covering the two-dimensional plane of the bridge is generated using ordinary kriging interpolation. The interpolation formula is as follows:

[0040] ;

[0041] In the formula, The interpolated damage index is the bridge's damage index at any coordinate (x, y) in the two-dimensional plane. This represents the number of neighboring sensors around the point to be interpolated. For the first Interpolation weighting coefficients for each sensor.

[0042] A system for implementing a bridge structural health analysis method based on vibration signals, the system comprising:

[0043] A vibration sensor array is deployed at multiple key monitoring nodes of the bridge structure to collect vibration signals of the bridge structure in real time.

[0044] The central processing unit, which is communicatively connected to the vibration sensor array, is configured to perform the following operations:

[0045] Receive and synchronize the vibration signals output by the vibration sensor array;

[0046] The aforementioned method steps include: selecting the master control node and assigning weights, generating a comprehensive feature vector using an adaptive filtering algorithm, constructing and evaluating the convergence of the damage distribution model, and generating a bridge structure damage distribution map.

[0047] The post-processing and visualization module is used to post-process the damage distribution map generated by the central processing unit and construct a three-dimensional visualization model containing bridge physical layout and damage distribution information.

[0048] Compared with the prior art, the present invention has the following beneficial effects:

[0049] 1. This invention effectively improves the signal-to-noise ratio and accuracy of damage identification through adaptive filtering and comprehensive feature vector extraction, and achieves near real-time analysis and feedback by combining convergence judgment.

[0050] 2. The damage distribution model established in this invention can simulate the dynamic evolution of damage, promptly capturing the generation and development of bridge structural damage. This overcomes the limitations of traditional static detection methods, achieving a leap from static detection to dynamic prediction. Furthermore, the damage distribution model is iteratively optimized through convergence indices until convergence, ensuring the accuracy and stability of the model's output results and providing a reliable technical basis for bridge maintenance decisions.

[0051] 3. By constructing a comprehensive feature vector, the vibration signal features are fully extracted. Combined with the damage index calculation formula, the degree of damage is quantitatively assessed. The ordinary Kriging interpolation algorithm is used to generate a two-dimensional damage distribution map of the entire bridge, so as to achieve accurate damage location and full-area visualization. Attached Figure Description

[0052] Figure 1 This is a schematic diagram of the method flow of the present invention;

[0053] Figure 2 This is a schematic diagram of the arrangement of the vibration sensor array;

[0054] Figure 3 This is a thermal diagram illustrating the structural health of a bridge. Detailed Implementation

[0055] The present invention will be further described below with reference to the accompanying drawings and embodiments, but this should not be construed as limiting the present invention.

[0056] Example: A method for bridge structural health analysis based on vibration signals, such as... Figure 1 As shown, it includes the following steps:

[0057] Step 1: Define and initialize the core parameters of the analysis system; deploy vibration sensor arrays at multiple key monitoring nodes of the bridge structure, and synchronize the output signals of the vibration sensor arrays to the central processing unit;

[0058] In this step, based on the bridge design parameters, service environment, and monitoring requirements, the following core parameters are preset. These parameters can be dynamically adjusted during implementation according to the actual situation:

[0059] Vibration signal sampling frequency: Set according to the vibration characteristics of the bridge. It is recommended to set it to 200Hz for conventional beam bridges and 500Hz for long-span bridges (span ≥ 500m).

[0060] Data resolution: ≥16 bits, ensuring the accuracy of vibration signal data;

[0061] Feature extraction window length: Set to 10~60s, select according to the dynamic load characteristics of the bridge, and take a smaller value for road sections with dense traffic.

[0062] Feedback control cycle: set to 5~30 minutes to achieve near real-time monitoring and analysis;

[0063] Environmental compensation coefficient: determined through preliminary calibration tests based on environmental factors such as temperature, humidity, and wind speed in the area where the bridge is located, with a range of 0.8 to 1.2;

[0064] Damage identification threshold: Determined through benchmark tests on bridges in a non-damaged state. The convergence index threshold is recommended to be set to 0.95, and the damage index threshold is set according to the characteristics of the bridge structural materials (0.3 for concrete structures and 0.25 for steel structures).

[0065] In this step, the vibration sensor array uses piezoelectric or fiber optic vibration sensors with an accuracy of not less than 0.001g and an adjustable sampling frequency (range ≥100Hz). The sensors must be able to resist environmental interference (temperature, humidity, electromagnetic interference) and be suitable for the outdoor service environment of bridges.

[0066] like Figure 2As shown, based on the bridge structure type (beam bridge, arch bridge, cable-stayed bridge, etc.) and stress characteristics, sensors are deployed at key locations where stress is concentrated and prone to damage, such as the mid-span of the main beam, the supports, the bottom of the bridge towers, and the anchorage ends of the cables. The node spacing is reasonably set according to the bridge span to ensure coverage of the entire bridge structure without any monitoring blind spots.

[0067] The central processing unit (CPU) utilizes an industrial-grade processor with multi-channel synchronous acquisition and high-speed computing capabilities. It supports wired / wireless (LoRa / 5G) communication with the vibration sensor array, with ≥16GB of processor memory and ≥512GB of storage capacity to meet the requirements for real-time vibration signal reception, processing, and data storage. A communication connection is established between the vibration sensor array and the CPU. Clock synchronization technology (such as the PTP precision clock protocol) ensures time synchronization of all sensor output signals, with a synchronization error ≤1ms, avoiding interference from signal time differences in subsequent analysis.

[0068] The deployment and signal synchronization process of the vibration sensor array is as follows: Based on the previously determined key monitoring nodes, the vibration sensors are installed and fixed on-site. The sensors must be installed in close contact with the bridge structure to avoid loose installation and additional vibration interference. The central processing unit and vibration sensor array are started, and the sensors undergo a power-on self-test to confirm that all sensors are working properly and signal transmission is smooth. Through the signal synchronization module of the central processing unit, the output signals of all sensors are synchronized to the central processing unit, establishing a one-to-one correspondence between sensor and node coordinates. A sensor deployment location information table is generated in the central processing unit, recording the sensor number, two-dimensional plane coordinates (x, y), and corresponding monitoring location for each sensor. The central processing unit activates the real-time signal acquisition mode, storing the acquired raw vibration signals in a standardized data format (such as CSV) and simultaneously displaying the signal acquisition status in real time for monitoring by technicians.

[0069] Step 2: Select a preset proportion of key sensors from the vibration sensor array as master control nodes and assign them initial weights;

[0070] In this step, the selection of master control nodes involves choosing 30% to 50% of the key sensors from the deployed vibration sensor array. The selection criteria are: monitoring locations of the bridge's core load-bearing components, high sensor signal-to-noise ratio, and strong data transmission stability. Priority is given to sensors located at critical locations such as the mid-span of the main beam and the supports. During the initial weight allocation process, a weighted method is used to assign initial weights to the master control nodes, with weight values ​​ranging from 0.6 to 0.9. The weights of the master control nodes are differentiated according to the importance of their monitoring locations (e.g., a weight of 0.9 for sensors at the mid-span of the main beam and 0.8 for sensors at the supports). The initial weights of other non-master control node sensors are uniformly set to 0.1 to 0.5. The initial weight values ​​of all sensors are entered and stored in the central processing unit.

[0071] Step 3: Based on the weights of the master control node and other sensors, the collected vibration signal is processed using an adaptive filtering algorithm, and a comprehensive feature vector is generated based on the filtered vibration signal.

[0072] This step uses an adaptive filtering algorithm to denoise and enhance the original vibration signal, and then generates a comprehensive feature vector through priority processing. The core is to retain the effective features related to structural damage in the vibration signal and remove interference signals such as environmental noise and traffic noise.

[0073] The adaptive filtering algorithm is executed as follows:

[0074] Load the adaptive filtering algorithm into the central processing unit and initialize the algorithm parameters: total number of sensors. The bias term represents the actual number of sensors deployed. Set the learning rate to 0.01. Set the value to 0.001~0.01 (adjust according to signal characteristics, take the smaller value when the signal-to-noise ratio is low), and set the target value of the filtered output to the reference vibration signal of the bridge under undamaged conditions.

[0075] The mathematical expression for the adaptive filtering algorithm is:

[0076] ;

[0077] In the formula: This is the filtered output signal. For time, For the first The weighting coefficients of each sensor; For the first The raw vibration signal collected by each sensor For bias terms, This represents the total number of sensors;

[0078] Based on the minimum mean square error criterion, the error between the filtered output and the target value is calculated in real time. And update the weight coefficients according to the following formula. :

[0079] ;

[0080] In the formula: For learning rate, To compensate for the error between the filtered output and the target value, the updated weight coefficients are fed back to the filtering calculation stage in real time to achieve adaptive filtering.

[0081] Furthermore, the process of generating the comprehensive feature vector is as follows:

[0082] Based on the initial weights assigned in step 2, and combined with the amplitude and signal-to-noise ratio of the vibration signals collected by the sensors, the contribution of each sensor is calculated. Then, the minimum-maximum normalization method is used to calculate the contribution. Normalized to the range of 0 to 1, the contribution is calculated as follows:

[0083] ;

[0084] In the formula, For the first The signal-to-noise ratio of each sensor signal.

[0085] The normalized contribution threshold is set to 0.5. Sensors with a contribution value greater than 0.5 are marked as high-priority nodes, and the rest are marked as low-priority nodes. A priority marking table is generated in the central processing unit.

[0086] The filtered vibration signals of high-priority nodes are processed using a weighted average method to generate enhanced feature vectors. The weights of the weighted average are the normalized contributions of each high-priority node, ensuring that the effective features of the core nodes are enhanced.

[0087] For the filtered vibration signal of low-priority nodes, wavelet denoising method (using db4 wavelet basis, decomposition level of 4) is used to remove interference signals such as environmental noise and traffic noise, extract effective features, and generate denoised feature vector.

[0088] The enhanced feature vector and the denoised feature vector are concatenated and merged according to the sensor node order to generate a comprehensive feature vector. The dimension of the comprehensive feature vector is consistent with the total number of sensors, thus fully preserving the effective vibration signal features of all sensors.

[0089] Step 4: Based on the comprehensive feature vector, construct a bridge structure damage distribution model to simulate the dynamic evolution of damage in the bridge structure;

[0090] In this step, the comprehensive feature vector generated in step 3 is used as input to construct a damage distribution model in the central processing unit to simulate the dynamic evolution of damage in the bridge structure. The mathematical expression of the model is as follows:

[0091] ;

[0092] In the formula: Let be the damage density distribution function. For time, The damage diffusion coefficient (determined based on the bridge structural materials; for concrete, D = 1 × 10⁻⁶) is given. −6 m 2 / s, steel structure D=5×10 −6 m 2 / s), For the Laplace operator, The source term represents the rate of damage generation;

[0093] Wherein, the source item The calculation relies on the comprehensive feature vector, which is used to transform the characteristics of the vibration signal into the driving force of damage generation. That is, the amplitude, rate of change, and other features of the comprehensive feature vector are used as the driving force of damage generation, and a linear fitting method is used to establish... With integrated feature vector Association model:

[0094]

[0095] In the formula, The fitting coefficients are determined through bridge damage tests, and b is a constant.

[0096] This enables the transformation of vibration signal characteristics into damage generation rate.

[0097] In this embodiment, the partial differential equations of the damage distribution model are numerically solved using the finite element method. The bridge structure is divided into several finite element elements, with each sensor node as an element node. By substituting the comprehensive feature vector and parameter values, the damage density at different times and spatial locations is solved. .

[0098] In this embodiment, to ensure the accuracy and stability of the damage distribution model's output, a convergence index is calculated and convergence is determined. If convergence fails, iterative processing is performed.

[0099] The formula for calculating the convergence index is as follows:

[0100] ;

[0101] In the formula: As a convergence index, For the first Each feature vector in time The value, The total number of eigenvectors;

[0102] The calculated convergence index Compare with a preset convergence threshold (recommended ≥0.95):

[0103] like ≥ Set a threshold, determine if the model has converged, and proceed to step 5, the damage distribution map generation stage;

[0104] like Set a threshold to determine if the model has not converged, and then use the current comprehensive feature vector. As new input, repeat the damage distribution model construction and solution process in this step until the convergence index reaches or exceeds the set threshold.

[0105] Step 5: After the damage distribution model converges, the comprehensive feature vector is mapped to a two-dimensional space to generate a bridge structure damage distribution map.

[0106] In this step, after the model converges, the integrated feature vectors are mapped onto the two-dimensional space of the bridge. A bridge structural damage distribution map (heatmap) is generated in four stages, achieving precise damage localization and full-domain visualization. All spatial analysis and plotting can be performed using ArcGIS / Matlab. Details are as follows:

[0107] Damage index extraction: Based on the comprehensive feature vector, the damage index corresponding to each sensor location is calculated. The damage index is a quantitative correlation value between the vibration signal characteristics and the degree of damage to the bridge structure, calculated based on the comprehensive feature vector, using the following formula:

[0108] ;

[0109] In the formula, For the first The damage index at each sensor monitoring location; the higher the value, the greater the degree of structural damage at that location. For the first The combined feature vector of each sensor, No. The baseline integrated feature vector of the bridge under undamaged conditions at each sensor location. , Both are weighting coefficients, and their sum is 1. They can be 0.7 and 0.3 respectively. For the first The standard deviation of the combined feature vector of each sensor characterizes the dispersion of the vibration signal; the higher the dispersion, the worse the structural stability.

[0110] Spatial coordinate association: Obtain the two-dimensional plane coordinates of each vibration sensor on the bridge structure; in this step, extract the two-dimensional plane coordinates (x, y) of each vibration sensor from the sensor deployment location information table established in step 1, and calculate the damage index of each sensor. By associating each sensor with its corresponding two-dimensional plane coordinates, a dataset is generated that associates sensor, damage index, and coordinates. The dataset format is (sensor number, x, y, ...). ).

[0111] Two-dimensional interpolation calculation: Based on the two-dimensional plane coordinates, the damage indicators are processed using a spatial interpolation algorithm to generate a continuous damage indicator distribution matrix covering the two-dimensional plane of the bridge. In this step, based on the two-dimensional plane coordinates of the sensor, a common Kriging interpolation algorithm is used to perform spatial interpolation processing on the damage indicators to generate a continuous damage indicator distribution matrix covering the entire two-dimensional plane of the bridge. The interpolation formula is:

[0112] ;

[0113] In the formula, The interpolated damage index is the bridge's damage index at any coordinate (x, y) in the two-dimensional plane. The number of neighboring sensors around the point to be interpolated, taken as 4-8. For the first The interpolation weighting coefficients of each sensor (which can be solved by the Kriging variance minimization criterion).

[0114] Generating a heatmap: The values ​​of the damage index distribution matrix are mapped to a preset color gradient to generate a heatmap of the bridge structure damage distribution. Specifically:

[0115] Color gradient preset: Establish a mapping relationship between damage indicators and color gradients, using a cool and warm color gradient system, such as: <0.2 (no damage) is set to blue, 0.2≤ <0.3 (minor damage) is set to yellow, 0.3≤ <0.5 (moderate damage) is set to orange. ≥0.5 (severe damage) is set to red, and the color depth increases as the damage index increases.

[0116] Heatmap creation: Import the damage index distribution matrix into a drawing tool. Using the two-dimensional planar layout of the bridge as the base map, map the interpolated damage index values ​​to a preset color gradient to generate a heatmap of bridge structural damage distribution. Mark the sensor node locations, damage index values, and damage levels on the heatmap. Add legends, coordinates, scale bars, and other elements to ensure the heatmap information is complete and intuitive. Figure 3 As shown.

[0117] Output Results: The central processing unit outputs and stores the generated damage distribution heat map in formats such as images (PNG / TIFF) and vector graphics (SVG). At the same time, it outputs damage index statistical reports, which include information such as damage index, damage level, and spatial location of each monitoring node, providing direct technical basis for bridge maintenance and repair decisions.

[0118] Example 2: This example provides a system for implementing the method described in Example 1. It is applicable to various large, medium, and small bridges, including beam bridges, arch bridges, and cable-stayed bridges. Through vibration signal acquisition, intelligent data processing, and three-dimensional visualization, it enables real-time monitoring, damage identification, and condition assessment of the bridge structure's health status. The system includes:

[0119] The vibration sensor array is deployed at key monitoring nodes prone to stress concentration and damage, such as the mid-span of the main girder, supports, bottom of the bridge towers, cable anchorage ends, and top of piers, based on the bridge structure type and stress characteristics. High-precision accelerometers are selected as the sensor type, and the sampling frequency can be adaptively adjusted within the range of 100-1000Hz. The spacing between sensor nodes is reasonably set according to the bridge span to ensure coverage of the entire bridge structure without monitoring blind spots. The sensor array establishes a stable connection with the central processing unit through wired or wireless communication to achieve real-time transmission of vibration signals.

[0120] The central processing unit, as the core control and data processing module of the system, communicates with the vibration sensor array. It has a built-in high-performance processor and dedicated data processing algorithms and is configured to perform the following operations:

[0121] 1. Signal reception and synchronization: Receives the raw vibration signals output by each vibration sensor and achieves precise synchronization of multi-sensor signals through timestamp synchronization technology, eliminating phase deviation caused by signal transmission delay;

[0122] 2. The implementation of bridge structural health analysis methods, including:

[0123] Master control node selection and weight allocation: Based on the signal quality, monitoring location importance and data stability of each sensor node, nodes with high signal-to-noise ratio and key monitoring coverage areas are automatically selected as master control nodes, and dynamic weight coefficients are allocated according to the monitoring priority of the nodes.

[0124] Comprehensive feature vector generation: The synchronized vibration signal is preprocessed (including denoising, filtering, and signal enhancement). An adaptive filtering algorithm is used to separate the effective signal from the interference signal. The time domain features (peak value, root mean square, kurtosis), frequency domain features (dominant frequency, spectral energy), and time-frequency domain features (wavelet packet energy entropy) of the signal are extracted and fused to generate a comprehensive feature vector.

[0125] Damage distribution model construction and convergence assessment: A bridge structure damage distribution model is constructed based on comprehensive feature vectors. The model parameters are adjusted through iterative optimization algorithms. At the same time, residual analysis, variance test and other methods are used to assess the convergence of the model to ensure the reliability of the model output results.

[0126] Damage distribution map generation: Based on the converged damage distribution model, the damage index of each monitoring node and the entire structure is calculated to generate a bridge structure damage distribution map.

[0127] 3. Data interaction and command issuance: The processing results are transmitted to the post-processing and visualization module, and external control commands can be received to adjust sensor sampling parameters, algorithm parameters, etc.

[0128] Post-processing and visualization module:

[0129] The system receives damage distribution maps from the central processing unit and performs post-processing operations such as smoothing, denoising, and boundary correction to eliminate outliers caused by data errors. Based on bridge design drawings and 3D modeling technology, it constructs a 3D visualization model that includes the bridge's physical layout (structural outlines such as main beams, towers, cables, supports, and piers) and damage distribution information (damage location, damage level, and damage index values). It supports interactive operations such as model scaling, rotation, and sectioning, and can intuitively distinguish different damage levels through color gradients. It also provides a damage information query function, allowing users to click on any location on the model to display detailed information such as damage indices and monitoring node data for the corresponding area.

[0130] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the claims.

Claims

1. A method for bridge structural health analysis based on vibration signals, characterized in that: Includes the following steps: Step 1: Deploy vibration sensor arrays at multiple key monitoring nodes of the bridge structure and synchronize the output signals of the vibration sensor arrays to the central processing unit; Step 2: Select a preset proportion of key sensors from the vibration sensor array as master control nodes and assign them initial weights; Step 3: Based on the weights of the master control node and other sensors, the collected vibration signal is processed using an adaptive filtering algorithm, and a comprehensive feature vector is generated based on the filtered vibration signal. Step 4: Based on the comprehensive feature vector, construct a bridge structure damage distribution model to simulate the dynamic evolution of damage in the bridge structure; Step 5: After the damage distribution model converges, the comprehensive feature vector is mapped to a two-dimensional space to generate a bridge structure damage distribution map.

2. The bridge structural health analysis method based on vibration signals according to claim 1, characterized in that: In step 1, the core parameters include at least the vibration signal sampling frequency, data resolution, feature extraction window length, feedback control cycle, environmental compensation coefficient, and damage identification threshold.

3. The method for bridge structural health analysis based on vibration signals according to claim 1, characterized in that: In step 3, the mathematical expression of the adaptive filtering algorithm is: ； In the formula: This is the filtered output signal. For time, For the first The weighting coefficients of each sensor; For the first The raw vibration signal collected by each sensor For bias terms, This represents the total number of sensors; The weighting coefficient The update is performed using the minimum mean square error criterion, and the update formula is as follows: ； In the formula: For learning rate, This represents the error between the filtered output and the target value.

4. The method for bridge structural health analysis based on vibration signals according to claim 1, characterized in that, Step 3, the step of generating the comprehensive feature vector, specifically includes: The contribution of each vibration sensor is calculated based on the initial weights, and the contribution is normalized to the range of 0 to 1. Sensors with a normalized contribution greater than 0.5 are marked as high-priority nodes, and the remaining sensors are marked as low-priority nodes. The vibration signals of high-priority nodes are processed by weighted averaging to generate enhanced feature vectors; The vibration signals of low-priority nodes are denoised to remove environmental noise and traffic noise interference, and a denoised feature vector is generated. The enhanced feature vector is merged with the denoised feature vector to generate the comprehensive feature vector.

5. The bridge structural health analysis method based on vibration signals according to claim 4, characterized in that: In step 4, the comprehensive feature vector is used to evaluate the convergence of the damage distribution model, and iterative processing is performed if convergence fails. The damage distribution model is used to simulate the dynamic evolution of damage in the bridge structure, and its mathematical model is as follows: ; In the formula: Let be the damage density distribution function. For time, The damage diffusion coefficient is... For the Laplace operator, The source term represents the damage generation rate. The calculation relies on the comprehensive feature vector, which is used to transform the characteristics of the vibration signal into the driving force for damage generation.

6. The method for bridge structural health analysis based on vibration signals according to claim 5, characterized in that: The convergence index of the damage distribution model is calculated based on the comprehensive feature vector, and the calculation formula is as follows: ； In the formula: As a convergence index, For the first Each feature vector in time The value, The total number of eigenvectors; If the convergence index is less than the set threshold, the construction steps of the repeated damage distribution model with the current comprehensive feature vector as input will continue until the convergence index reaches or exceeds the set threshold.

7. The method for bridge structural health analysis based on vibration signals according to claim 5, characterized in that: The step of mapping feature vectors to a two-dimensional space to generate a bridge structure damage distribution map includes: Damage index extraction: Based on the comprehensive feature vector, calculate the damage index corresponding to each sensor location; Spatial coordinate correlation: Obtain the two-dimensional planar coordinates of each vibration sensor on the bridge structure; Two-dimensional interpolation calculation: Based on the two-dimensional plane coordinates, the damage index is processed using a spatial interpolation algorithm to generate a continuous damage index distribution matrix covering the two-dimensional plane of the bridge; Generating a heat map: The values ​​of the damage index distribution matrix are mapped to a preset color gradient to generate a heat map of the bridge structure damage distribution.

8. The method for bridge structural health analysis based on vibration signals according to claim 7, characterized in that: The damage index is a quantitative correlation value between vibration signal characteristics and the degree of damage to the bridge structure. It is calculated based on a comprehensive feature vector, and the formula is as follows: ； In the formula, For the first The damage index at each sensor monitoring location; the higher the value, the greater the degree of structural damage at that location. For the first The combined feature vector of each sensor, No. The baseline integrated feature vector of the bridge under undamaged conditions at each sensor location. , Both are weighting coefficients, and their sum is 1; For the first The standard deviation of the combined feature vector of each sensor characterizes the dispersion of the vibration signal; the higher the dispersion, the worse the structural stability.

9. The bridge structural health analysis method based on vibration signals according to claim 8, characterized in that: Based on damage indicators A continuous damage index distribution matrix covering the two-dimensional plane of the bridge is generated using ordinary kriging interpolation. The interpolation formula is as follows: ； In the formula, The interpolated damage index is the bridge's damage index at any coordinate (x, y) in the two-dimensional plane. This represents the number of neighboring sensors around the point to be interpolated. For the first Interpolation weighting coefficients for each sensor.

10. A system for implementing a bridge structural health analysis method based on vibration signals, characterized in that, The system includes: A vibration sensor array is deployed at multiple key monitoring nodes of the bridge structure to collect vibration signals of the bridge structure in real time. The central processing unit, which is communicatively connected to the vibration sensor array, is configured to perform the following operations: Receive and synchronize the vibration signals output by the vibration sensor array; The method steps of any one of claims 1 to 9 include: selecting a master control node and assigning weights, generating a comprehensive feature vector using an adaptive filtering algorithm, constructing and evaluating the convergence of the damage distribution model, and generating a bridge structure damage distribution map; The post-processing and visualization module is used to post-process the damage distribution map generated by the central processing unit and construct a three-dimensional visualization model containing bridge physical layout and damage distribution information.

Images