Bridge driving early warning system and method for intelligent transportation

Abstract

This application discloses a bridge traffic early warning system and method for intelligent transportation, belonging to the field of intelligent transportation and bridge structural health monitoring technology. The method includes: calculating a single load event risk score; constructing a short-term analysis window, performing cumulative damage calculation and trend analysis on the risk score sequence within the short-term analysis window to obtain a short-term trend risk score; calculating a structural health index and dynamically correcting a preset risk judgment threshold; weightedly fusing the single load event risk score, the short-term trend risk score, and the structural health index to obtain a dynamic risk index; determining whether the dynamic risk index exceeds the corrected risk judgment threshold: if it exceeds, a graded early warning judgment is made based on the dynamic risk index, triggering the corresponding level of early warning signal and implementing corresponding traffic control measures; if it does not exceed, the current traffic operation status is maintained, and bridge structural response data and traffic load data continue to be collected.

Description

Technical Field

[0001] This application relates to the field of intelligent transportation and bridge structural health monitoring technology, and more specifically, to an intelligent transportation bridge driving early warning system and method. Background Technology

[0002] With the development of intelligent transportation and digital infrastructure, bridges, as key nodes in transportation networks, are directly related to road traffic efficiency and public safety. In recent years, with the continuous growth of traffic flow and the increasing proportion of heavy-duty vehicles, the load levels and frequency of bridges during their service life have increased significantly, leading to increasingly prominent issues of structural performance degradation and sudden risks. Therefore, building a technical system capable of real-time perception, dynamic assessment, and proactive early warning of bridge operating status has become an important research direction in the field of intelligent transportation.

[0003] Existing bridge safety monitoring technologies largely rely on structural health monitoring systems. These systems use sensors to acquire response data such as strain, acceleration, and displacement, and then assess the bridge's condition based on threshold judgments or empirical models. However, these methods typically focus on analyzing long-term structural performance changes and lack the ability to identify transient risks caused by short-term traffic loads, making it difficult to promptly reflect dynamic safety hazards during vehicle traffic. Furthermore, traditional methods often use fixed thresholds for risk assessment, failing to fully consider the impact of environmental factors (such as temperature changes) and structural performance evolution on response characteristics, which can easily lead to false alarms or missed alarms.

[0004] On the other hand, with the development of data-driven methods, some studies have attempted to introduce machine learning models to model bridge response and load characteristics in order to improve the accuracy of risk assessment. However, existing methods often only analyze data based on a single dimension, such as a single load response or statistical characteristics, lacking comprehensive integration of multi-source information (such as single event risk, short-term cumulative effects, and overall structural health status), making it difficult to fully characterize the true risk level of bridges in complex traffic environments. Furthermore, for complex conditions such as continuous multi-vehicle traffic and simultaneous loading of two-way lanes, existing technologies still have shortcomings in event segmentation, risk superposition, and evolution trend identification.

[0005] Furthermore, regarding early warning mechanisms, existing technologies typically employ static hierarchical strategies, lacking the ability to adaptively adjust early warning thresholds based on the bridge's current health status. This leads to discrepancies between early warning results and the actual structural safety margin. Simultaneously, the linkage mechanism between early warning results and traffic control measures is inadequate, hindering the achievement of closed-loop management from risk identification to proactive intervention, thus limiting the system's effectiveness in practical engineering applications.

[0006] In summary, how to achieve multi-dimensional dynamic assessment of bridge operation risks based on the integration of multi-source monitoring data, and how to adaptively adjust the early warning thresholds in conjunction with the structural health status, thereby realizing accurate and real-time hierarchical early warning and effective traffic control linkage, has become an urgent technical problem to be solved. Summary of the Invention

[0007] In order to overcome a series of defects in the existing technology, the purpose of this application is to provide a bridge driving warning method for intelligent transportation, which includes the following steps:

[0008] Based on bridge structural response data and traffic load data, the net structural response signal is obtained;

[0009] Load event detection and interval division are performed based on the net structural response signal, and the risk score of a single load event is calculated.

[0010] Construct a short-term analysis window, perform cumulative damage calculation and trend analysis on the risk score sequence within the short-term analysis window, and obtain a short-term trend risk score;

[0011] Statistical analysis and modal feature extraction of the net structural response signal are performed based on a fixed sliding time window, the structural health index is calculated, and the preset risk judgment threshold is dynamically corrected based on the structural health index.

[0012] The dynamic risk index is obtained by weighting and fusing the single load event risk score, the short-term trend risk score and the structural health index.

[0013] Determine whether the dynamic risk index exceeds the revised risk assessment threshold:

[0014] If the risk level is exceeded, a graded early warning will be issued based on the dynamic risk index, triggering the corresponding level of early warning signal and implementing the corresponding traffic control measures.

[0015] If the limit is not exceeded, the current traffic operation status will be maintained, and bridge structural response data and traffic load data will continue to be collected for the next round of testing and evaluation.

[0016] In some embodiments, the method for obtaining the net structure response signal

[0017] The bridge structural response data and traffic load data are filtered to obtain a smooth signal;

[0018] Based on data from periods of low load or no vehicle traffic, an environmental background baseline for the bridge structure is obtained by fitting.

[0019] The environmental background baseline is subtracted from the smoothed signal to obtain the structural response signal with the environmental background removed;

[0020] Based on key section temperature data and structural response signals after removing environmental background, the temperature influence components of the bridge structure are identified.

[0021] The net structural response signal is obtained by subtracting the temperature component from the structural response signal after removing the influence of the environmental background.

[0022] In some embodiments, the method for performing load event detection and interval division is as follows:

[0023] The start and end times of the load event are determined based on the net structural response signal and the axle load information obtained by the vehicle detection equipment.

[0024] The analysis interval for a single load event is constructed by using the preset duration before the start time and the preset duration after the end time as buffer boundaries.

[0025] When the time interval between two adjacent load events is less than a preset interval threshold, the adjacent load events are merged into consecutive load events, and the number of merged events is recorded.

[0026] When the duration of the continuous load event exceeds a preset duration threshold, the continuous load event is segmented based on the local minimum point of the net structural response signal, and the duration of each segment is not less than the preset minimum duration.

[0027] In some embodiments, the method for calculating the risk score of a single load event is as follows:

[0028] Based on the defined load event analysis interval, load event characteristic quantities are extracted, including the net structural response peak value, the ratio of the peak value to the bridge design bearing capacity response benchmark value, the response rise slope, the response duration, the response integral area, the interval between adjacent load events, the estimated total axle load of vehicles passing at the same time, and the multi-lane synchronous loading marker.

[0029] The load event features are input into a pre-trained gradient boosting regression model to perform risk mapping calculations and obtain a risk score for a single load event.

[0030] In some embodiments, the method for constructing a short-term analysis window is as follows:

[0031] The frequency of load events is statistically analyzed based on a fixed time step, and the statistical features within each time step are obtained and stored in a counting buffer. The statistical features include: the total number of load events, the number of high-risk events, and the number of simultaneous over-limit events in both directions of the lane.

[0032] Based on the cumulative statistical results within three consecutive fixed time steps, a condition judgment is made. When the total number of cumulative load events is not less than the preset number threshold, or the number of high-risk events is not less than the preset risk number threshold, or the number of bidirectional synchronization over-limit events is not less than the preset synchronization number threshold, the construction of a short-term analysis window is triggered.

[0033] When the construction conditions are triggered, the current time is used as the end point of the window, and three fixed time steps are traced back as the starting point of the window to form a short-term analysis window.

[0034] In some embodiments, the method for obtaining the short-term trend risk score is as follows:

[0035] Equivalent stress mapping is performed on the risk scores of single load events arranged in time series within the short-time analysis window, and the corresponding SN curve parameters are selected based on the bridge structure type and material parameters to obtain the equivalent stress amplitude sequence.

[0036] The equivalent stress amplitude sequence is divided into hierarchical intervals to construct stress amplitude distribution intervals, and the number of cycles within each distribution interval is counted.

[0037] Based on the number of cycles in each stress amplitude range and the allowable number of cycles for the SN curve in the corresponding range, the fatigue damage increment in the current short-time analysis window is calculated according to the linear cumulative damage criterion.

[0038] The fatigue damage increment is superimposed with the historical cumulative damage value to obtain the cumulative damage index characterizing the degree of structural deterioration.

[0039] Trend fitting analysis is performed on the risk scores of single load events arranged in time series within the short-term analysis window, and the trend slope index and trend direction indicator are adaptively determined based on the goodness of fit.

[0040] The coefficient of variation is calculated based on the risk score sequence of a single load event within a short-time analysis window.

[0041] The fatigue damage increment, trend slope index, and coefficient of variation are used as input parameters for fusion calculation. Under the condition of satisfying the normalization constraint of weight coefficient, the short-term trend risk score is obtained.

[0042] In some embodiments, the method for statistical analysis and modal feature extraction of the net structural response signal based on a fixed sliding time window is as follows:

[0043] A fixed sliding time window is constructed, and the window length and sliding step size are set. The net structural response signal is segmented to obtain a window data sequence arranged in time series.

[0044] Within each fixed sliding time window, statistical characteristics of the net structural response signal are calculated to obtain statistical parameters including mean, standard deviation, peak value, peak-to-valley difference, skewness, and kurtosis.

[0045] Calculate the rate of change of statistical parameters based on adjacent sliding time windows;

[0046] Within each sliding time window, modal identification processing is performed on the net structural response signal to extract multi-mode parameters of the structure, including natural frequency, damping ratio and mode participation coefficient.

[0047] Based on a preset stability criterion, modal parameters are screened to obtain a set of stable modal parameters;

[0048] By comparing and analyzing the stable modal parameters with the preset deviation criteria, the natural frequency deviation, damping ratio deviation, and modal confidence criterion values ​​for each order are obtained.

[0049] In some embodiments, the method for calculating the structural health index is as follows:

[0050] The rate of change of statistical parameters, the frequency deviations of each order, the damping ratio deviations, and the modal confidence criterion values ​​are collected to obtain a multi-source feature input vector;

[0051] The membership degree transformation process is performed on each feature quantity in the multi-source feature input vector, and the membership degree value corresponding to each feature quantity is obtained based on the preset trapezoidal membership degree function.

[0052] The comprehensive evaluation result is obtained by performing a weighted summation based on the membership values ​​corresponding to each feature and the preset weight vector.

[0053] The comprehensive evaluation results are normalized to obtain the structural health index.

[0054] In some embodiments, the dynamic risk index is obtained as follows:

[0055] The basic weighting coefficients for the single load event risk score, short-term trend risk score, and structural health index are set to 0.4, 0.4, and 0.2, respectively.

[0056] The basic weighting coefficients are adaptively adjusted based on real-time data quality: when the confidence interval width of a single load event risk score exceeds 15 points, its weighting coefficient is reduced by 0.05, and the corresponding increment is evenly distributed to the short-term trend risk score and the structural health index; when the number of effective load events within the short-term analysis window is less than 5, the weighting coefficient of the short-term trend risk score is reduced by 0.05, and the corresponding increment is evenly distributed to the single load event risk score; when the standard deviation of the structural health index's three most recent update values ​​exceeds 10, its weighting coefficient is reduced by 0.05, and the corresponding increment is evenly distributed to the single load event risk score and the short-term trend risk score.

[0057] The adaptively adjusted weight coefficients are corrected so that the sum of the weight coefficients satisfies the normalization constraint and that no weight coefficient is lower than 0.10.

[0058] Based on the corrected weighting coefficients, the single load event risk score, short-term trend risk score, and structural health index are weighted and linearly combined to obtain the dynamic risk index.

[0059] The purpose of this application is also to provide an intelligent traffic bridge driving warning system for implementing the above-mentioned bridge driving warning method, including:

[0060] The data acquisition module is used to collect bridge structural response data and traffic load data in real time.

[0061] The signal processing module is used to acquire the net structural response signal of the bridge structure;

[0062] The event detection module is used to realize the real-time identification and classification of load events;

[0063] The damage assessment module is used to quantify the degree of cumulative damage and development trend of bridges in a short period of time.

[0064] The health monitoring module is used to comprehensively assess the overall condition and long-term safety performance of the bridge structure.

[0065] The early warning and control module is responsible for triggering early warning signals of corresponding levels to achieve proactive management of bridge safety operation.

[0066] Compared with the prior art, this application has the following beneficial effects:

[0067] This application achieves accurate assessment and adaptive graded early warning of bridge operation risks by integrating single load risk, short-term trend risk, and structural health status, and dynamically correcting the early warning threshold. This significantly improves the real-time performance and accuracy of bridge safety monitoring and the effectiveness of traffic control. Attached Figure Description

[0068] Figure 1 This is a flowchart illustrating a bridge driving warning method for intelligent transportation disclosed in an embodiment of this application.

[0069] Figure 2 This is a schematic diagram of the architecture of a bridge traffic warning system for intelligent transportation disclosed in an embodiment of this application. Detailed Implementation

[0070] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of the embodiments of this invention will be described in more detail below with reference to the accompanying drawings. In the drawings, the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The described embodiments are some embodiments of this invention, but not all embodiments.

[0071] Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0072] The embodiments and directional terms described below with reference to the accompanying drawings are exemplary and intended to explain the present invention, and should not be construed as limiting the present invention.

[0073] In a broad embodiment of this application, a bridge driving warning method for intelligent transportation includes the following steps:

[0074] Based on bridge structural response data and traffic load data, the net structural response signal is obtained;

[0075] Load event detection and interval division are performed based on the net structural response signal, and the risk score of a single load event is calculated.

[0076] Construct a short-term analysis window, perform cumulative damage calculation and trend analysis on the risk score sequence within the short-term analysis window, and obtain a short-term trend risk score;

[0077] Statistical analysis and modal feature extraction of the net structural response signal are performed based on a fixed sliding time window, the structural health index is calculated, and the preset risk judgment threshold is dynamically corrected based on the structural health index.

[0078] The dynamic risk index is obtained by weighting and fusing the single load event risk score, the short-term trend risk score and the structural health index.

[0079] Determine whether the dynamic risk index exceeds the revised risk assessment threshold:

[0080] If the risk level is exceeded, a graded early warning will be issued based on the dynamic risk index, triggering the corresponding level of early warning signal and implementing the corresponding traffic control measures.

[0081] If the limit is not exceeded, the current traffic operation status will be maintained, and bridge structural response data and traffic load data will continue to be collected for the next round of testing and evaluation.

[0082] In some embodiments, the method for obtaining the net structure response signal

[0083] The bridge structural response data and traffic load data are filtered to obtain a smooth signal;

[0084] Based on data from periods of low load or no vehicle traffic, an environmental background baseline for the bridge structure is obtained by fitting.

[0085] The environmental background baseline is subtracted from the smoothed signal to obtain the structural response signal with the environmental background removed;

[0086] Based on key section temperature data and structural response signals after removing environmental background, the temperature influence components of the bridge structure are identified.

[0087] The net structural response signal is obtained by subtracting the temperature component from the structural response signal after removing the influence of the environmental background.

[0088] In some embodiments, the method for performing load event detection and interval division is as follows:

[0089] The start and end times of the load event are determined based on the net structural response signal and the axle load information obtained by the vehicle detection equipment.

[0090] The analysis interval for a single load event is constructed by using the preset duration before the start time and the preset duration after the end time as buffer boundaries.

[0091] When the time interval between two adjacent load events is less than a preset interval threshold, the adjacent load events are merged into consecutive load events, and the number of merged events is recorded.

[0092] When the duration of the continuous load event exceeds a preset duration threshold, the continuous load event is segmented based on the local minimum point of the net structural response signal, and the duration of each segment is not less than the preset minimum duration.

[0093] In some embodiments, the method for calculating the risk score of a single load event is as follows:

[0094] Based on the defined load event analysis interval, load event characteristic quantities are extracted, including the net structural response peak value, the ratio of the peak value to the bridge design bearing capacity response benchmark value, the response rise slope, the response duration, the response integral area, the interval between adjacent load events, the estimated total axle load of vehicles passing at the same time, and the multi-lane synchronous loading marker.

[0095] The load event features are input into a pre-trained gradient boosting regression model to perform risk mapping calculations and obtain a risk score for a single load event.

[0096] In some embodiments, the method for constructing a short-term analysis window is as follows:

[0097] The frequency of load events is statistically analyzed based on a fixed time step, and the statistical features within each time step are obtained and stored in a counting buffer. The statistical features include: the total number of load events, the number of high-risk events, and the number of simultaneous over-limit events in both directions of the lane.

[0098] Based on the cumulative statistical results within three consecutive fixed time steps, a condition judgment is made. When the total number of cumulative load events is not less than the preset number threshold, or the number of high-risk events is not less than the preset risk number threshold, or the number of bidirectional synchronization over-limit events is not less than the preset synchronization number threshold, the construction of a short-term analysis window is triggered.

[0099] When the construction conditions are triggered, the current time is used as the end point of the window, and three fixed time steps are traced back as the starting point of the window to form a short-term analysis window.

[0100] In some embodiments, the method for obtaining the short-term trend risk score is as follows:

[0101] Equivalent stress mapping is performed on the risk scores of single load events arranged in time series within the short-time analysis window, and the corresponding SN curve parameters are selected based on the bridge structure type and material parameters to obtain the equivalent stress amplitude sequence.

[0102] The equivalent stress amplitude sequence is divided into hierarchical intervals to construct stress amplitude distribution intervals, and the number of cycles within each distribution interval is counted.

[0103] Based on the number of cycles in each stress amplitude range and the allowable number of cycles for the SN curve in the corresponding range, the fatigue damage increment in the current short-time analysis window is calculated according to the linear cumulative damage criterion.

[0104] The fatigue damage increment is superimposed with the historical cumulative damage value to obtain the cumulative damage index characterizing the degree of structural deterioration.

[0105] Trend fitting analysis is performed on the risk scores of single load events arranged in time series within the short-term analysis window, and the trend slope index and trend direction indicator are adaptively determined based on the goodness of fit.

[0106] The coefficient of variation is calculated based on the risk score sequence of a single load event within a short-time analysis window.

[0107] The fatigue damage increment, trend slope index, and coefficient of variation are used as input parameters for fusion calculation. Under the condition of satisfying the normalization constraint of weight coefficient, the short-term trend risk score is obtained.

[0108] In some embodiments, the method for statistical analysis and modal feature extraction of the net structural response signal based on a fixed sliding time window is as follows:

[0109] A fixed sliding time window is constructed, and the window length and sliding step size are set. The net structural response signal is segmented to obtain a window data sequence arranged in time series.

[0110] Within each fixed sliding time window, statistical characteristics of the net structural response signal are calculated to obtain statistical parameters including mean, standard deviation, peak value, peak-to-valley difference, skewness, and kurtosis.

[0111] Calculate the rate of change of statistical parameters based on adjacent sliding time windows;

[0112] Within each sliding time window, modal identification processing is performed on the net structural response signal to extract multi-mode parameters of the structure, including natural frequency, damping ratio and mode participation coefficient.

[0113] Based on a preset stability criterion, modal parameters are screened to obtain a set of stable modal parameters;

[0114] By comparing and analyzing the stable modal parameters with the preset deviation criteria, the natural frequency deviation, damping ratio deviation, and modal confidence criterion values ​​for each order are obtained.

[0115] In some embodiments, the method for calculating the structural health index is as follows:

[0116] The rate of change of statistical parameters, the frequency deviations of each order, the damping ratio deviations, and the modal confidence criterion values ​​are collected to obtain a multi-source feature input vector;

[0117] The membership degree transformation process is performed on each feature quantity in the multi-source feature input vector, and the membership degree value corresponding to each feature quantity is obtained based on the preset trapezoidal membership degree function.

[0118] The comprehensive evaluation result is obtained by performing a weighted summation based on the membership values ​​corresponding to each feature and the preset weight vector.

[0119] The comprehensive evaluation results are normalized to obtain the structural health index.

[0120] In some embodiments, the dynamic risk index is obtained as follows:

[0121] The basic weighting coefficients for the single load event risk score, short-term trend risk score, and structural health index are set to 0.4, 0.4, and 0.2, respectively.

[0122] The basic weighting coefficients are adaptively adjusted based on real-time data quality: when the confidence interval width of a single load event risk score exceeds 15 points, its weighting coefficient is reduced by 0.05, and the corresponding increment is evenly distributed to the short-term trend risk score and the structural health index; when the number of effective load events within the short-term analysis window is less than 5, the weighting coefficient of the short-term trend risk score is reduced by 0.05, and the corresponding increment is evenly distributed to the single load event risk score; when the standard deviation of the structural health index's three most recent update values ​​exceeds 10, its weighting coefficient is reduced by 0.05, and the corresponding increment is evenly distributed to the single load event risk score and the short-term trend risk score.

[0123] The adaptively adjusted weight coefficients are corrected so that the sum of the weight coefficients satisfies the normalization constraint and that no weight coefficient is lower than 0.10.

[0124] Based on the corrected weighting coefficients, the single load event risk score, short-term trend risk score, and structural health index are weighted and linearly combined to obtain the dynamic risk index.

[0125] The purpose of this application is also to provide an intelligent traffic bridge driving warning system for implementing the above-mentioned bridge driving warning method, including:

[0126] The data acquisition module is used to collect bridge structural response data and traffic load data in real time.

[0127] The signal processing module is used to acquire the net structural response signal of the bridge structure;

[0128] The event detection module is used to realize the real-time identification and classification of load events;

[0129] The damage assessment module is used to quantify the degree of cumulative damage and development trend of bridges in a short period of time.

[0130] The health monitoring module is used to comprehensively assess the overall condition and long-term safety performance of the bridge structure.

[0131] The early warning and control module is responsible for triggering early warning signals of corresponding levels to achieve proactive management of bridge safety operation.

[0132] The preferred embodiments of this application are described below with reference to the accompanying drawings, and the application will be further described in detail.

[0133] Figure 1 This is a schematic flowchart illustrating a bridge traffic warning method for intelligent transportation, provided in the first aspect of this application. The method is applicable to bridge operation status monitoring and traffic safety management scenarios. The method includes the following steps.

[0134] S1, based on bridge structural response data and traffic load data, yields the net structural response signal.

[0135] The method for obtaining the net structural response signal includes the following steps:

[0136] S11 filters the bridge structure response data and traffic load data to obtain a smooth signal.

[0137] Specifically, bridge structural response data (including strain, acceleration, or displacement) and traffic load data (including vehicle axle load and traffic information) are collected in real time. The data is then subjected to noise reduction filtering to eliminate high-frequency noise and abnormal fluctuations, resulting in a continuous and stable smooth signal.

[0138] For example, the filtering process may employ low-pass filtering, moving average filtering, or Kalman filtering methods to improve the signal-to-noise ratio and ensure the stability of subsequent analysis.

[0139] S12, based on data from periods of low load or no vehicle traffic, is used to fit the environmental background baseline of the bridge structure.

[0140] Specifically, time periods with low loads or no vehicle traffic are selected from historical data, and the corresponding structural response signals are extracted as environmental response samples. Based on the environmental response samples, an environmental background baseline model is established using regression analysis or curve fitting methods to characterize the influence of environmental factors such as temperature and humidity on the structural response.

[0141] For example, the fitting method may include polynomial fitting, spline function fitting, or time series-based trend decomposition methods.

[0142] S13, subtract the environmental background baseline from the smoothed signal to obtain the structural response signal with the environmental background removed.

[0143] Specifically, the environmental background baseline is used as a reference value, and the baseline value at the corresponding time is subtracted point by point from the smoothed signal to eliminate the low-frequency variation components caused by environmental factors, so as to obtain a structural response signal that only reflects the effect of traffic load.

[0144] S14. Based on key section temperature data and structural response signals after removing environmental background, the temperature influence component of the bridge structure is identified.

[0145] Specifically, temperature monitoring data of key bridge sections are obtained and correlated with the structural response signal after removing the environmental background is analyzed or regression modeled to identify the coupled effect of temperature changes on the structural response and extract the temperature influence component.

[0146] For example, linear regression or multivariate regression models can be used to model the relationship between temperature and structural response to improve the accuracy of temperature effect identification.

[0147] S15, subtract the temperature influence component from the structural response signal after removing the influence of the environmental background to obtain the net structural response signal.

[0148] Specifically, the identified temperature influence component is further subtracted from the structural response signal after removing the environmental background, thereby obtaining the net structural response signal caused only by traffic load, which is used for subsequent load event detection and risk assessment analysis.

[0149] The method for obtaining the net structural response signal described in this application effectively extracts the true structural response under traffic load by filtering the bridge structural response data, subtracting the environmental background baseline, and removing the temperature influence component. This significantly reduces the interference of environmental factors on the monitoring results, improves the accuracy and interpretability of the structural response data, and thus provides a reliable data foundation for subsequent load event identification and risk assessment.

[0150] S2 performs load event detection and interval division based on the net structural response signal, and calculates the risk score of a single load event.

[0151] The methods for load event detection, interval division, and single load event risk score calculation include the following steps:

[0152] S21. Based on the net structural response signal and the axle load information obtained by the vehicle detection equipment, determine the start and end times of the load event.

[0153] Specifically, the system acquires the bridge's net structural response signal and the axle load information output by the vehicle detection equipment in real time, and performs synchronous alignment processing on the signals; based on the changes in structural response amplitude and the axle load trigger signal, it identifies the response mutation interval during the load application process, thereby determining the start and end times of the load event.

[0154] For example, by setting a response amplitude threshold or a rate of change threshold, combined with the trigger flag of the axle load signal, accurate identification of single-vehicle or multi-vehicle load events can be achieved.

[0155] S22, using the preset duration before the start time and the preset duration after the end time as buffer boundaries, construct the analysis interval for a single load event.

[0156] Specifically, based on the determined start and end times of the load events, the preset time lengths are extended forward and backward respectively to form an analysis interval that includes the complete loading and unloading process, so as to avoid the impact of boundary truncation on feature extraction.

[0157] For example, the preset duration can be set according to the dynamic response characteristics of the bridge structure.

[0158] S23, when the time interval between two adjacent load events is less than the preset interval threshold, the adjacent load events are merged into consecutive load events, and the number of merged events is recorded.

[0159] Specifically, the time interval between adjacent load events is calculated. When the interval is less than a preset interval threshold, the corresponding events are merged into consecutive load events to characterize dense vehicle traffic or convoy loading. The number of merged events is then recorded.

[0160] S24, when the duration of the continuous load event exceeds a preset duration threshold, the continuous load event is segmented based on the local minimum point of the net structural response signal.

[0161] Specifically, when the duration of a continuous load event is long, the event is segmented based on the local minimum points in the net structural response signal to distinguish different vehicles or loading sub-processes; and the duration of a single segment after segmentation is not less than the preset minimum duration to ensure the effectiveness and physical meaning of the segmentation results.

[0162] S25. Based on the defined load event analysis interval, extract the load event characteristic quantities.

[0163] Specifically, within each single load event analysis interval, multidimensional features characterizing the structural response are extracted, including the net structural response peak value, the ratio of the peak value to the bridge design bearing capacity response benchmark value, the slope of the response rise edge, the duration of the response, the area of ​​the response integral, the interval between adjacent load events, the estimated total axle load of vehicles passing in the same period, and the multi-lane synchronous loading marker.

[0164] For example, the multi-lane synchronous loading flag is used to reflect the adverse effects of multiple lanes being loaded simultaneously on the bridge structure.

[0165] S26. Input the load event features into the pre-trained gradient boosting regression model to perform risk mapping calculation and obtain the risk score for a single load event.

[0166] Specifically, the extracted load event features are used as input vectors and fed into a pre-trained gradient boosting regression model to output the corresponding risk score. The risk score ranges from 0 to 100 and is used to characterize the risk level of the load event on the bridge structure. The higher the score, the higher the risk level.

[0167] For example, when the score is close to 100, it indicates that the structural response has approached or reached the design ultimate bearing capacity.

[0168] S27 outputs the risk score for a single load event and provides a confidence interval estimate.

[0169] Specifically, based on the prediction results of the gradient boosting regression model, a single load event risk score is output, and the corresponding confidence interval is calculated to characterize the uncertainty of the prediction results. When the confidence interval width exceeds a preset threshold (e.g., 15 points), an uncertainty marker is added to the risk score to indicate that the reliability of the result has decreased.

[0170] For example, the training samples of the gradient boosting regression model are derived from a mixed dataset consisting of historical measured data and finite element simulation data. The number of training samples is no less than 5,000, and the root mean square error of the validation set does not exceed 3 points, thereby ensuring the model's prediction accuracy and generalization ability.

[0171] The load event detection and single-event risk assessment method described in this application achieves accurate identification and reasonable interval division of load events by integrating net structural response signals and axle load information. Based on this, multidimensional feature quantities are extracted and risk mapping calculations are completed by combining gradient boosting regression models. At the same time, confidence intervals are introduced to characterize the uncertainty of the results, thereby achieving accurate quantification of the risk of single load events under complex traffic conditions, providing a reliable foundation for subsequent short-term trend analysis and comprehensive risk assessment.

[0172] S3. Construct a short-term analysis window, perform cumulative damage calculation and trend analysis on the risk score sequence within the short-term analysis window, and obtain the short-term trend risk score.

[0173] The method for constructing a short-term analysis window and calculating a short-term trend risk score includes the following steps:

[0174] S31, based on a fixed time step, the frequency of load events is statistically analyzed to obtain statistical features and stored in a counting buffer.

[0175] Specifically, the identified load events are segmented and statistically analyzed according to a preset fixed time step. Within each time step, the total number of load events, the number of high-risk events, and the number of simultaneous overload events in both directions are calculated. These statistical features are then stored in a counting buffer to form time series statistical data.

[0176] For example, the fixed time step can be set according to traffic flow characteristics to balance statistical stability and real-time response.

[0177] S32, based on the cumulative statistical results of multiple consecutive time steps, makes a conditional judgment and triggers the construction of a short-term analysis window.

[0178] Specifically, the statistical characteristics within three consecutive fixed time steps are cumulatively calculated. When the total number of cumulative load events is not less than the preset threshold, or the number of high-risk events is not less than the preset risk threshold, or the number of bidirectional synchronization over-limit events is not less than the preset synchronization threshold, the triggering condition is met, and the short-term analysis window construction process is started.

[0179] S33, when the triggering condition is met, determine the start and end range of the short-term analysis window.

[0180] Specifically, the current moment is used as the end point of the window, and three consecutive fixed time steps are traced back as the starting point of the window to construct a short-term analysis window that covers recent high-risk load behavior.

[0181] S34 processes missing data within the short-term analysis window.

[0182] Specifically, for missing data segments caused by data acquisition interruption, when the cumulative duration of missing data does not exceed a preset proportion threshold of the total window duration, an interpolation method is used to complete the data; when the duration of missing data exceeds the proportion threshold, the short-term analysis window is extended forward until the data integrity requirements are met or the preset maximum number of extensions is reached.

[0183] S35 records window metadata and establishes associations.

[0184] Specifically, the start and end times of the short-term analysis window and the total number of load events within the window are recorded as window metadata. The window metadata is then linked to the short-term trend risk score calculated subsequently to facilitate historical tracing and status analysis.

[0185] S36 performs a sliding update of the short-term analysis window and calculates the risk evolution rate.

[0186] Specifically, when the short-term analysis window is triggered multiple times in succession, the window is updated according to the preset sliding step size, and the change between two adjacent short-term trend risk scores is calculated to characterize the rate of risk evolution and serve as a reference indicator for subsequent comprehensive assessment.

[0187] S37 performs equivalent stress mapping on the risk score sequence within the short-term analysis window.

[0188] Specifically, the risk scores of single load events arranged in time sequence within the window are mapped to equivalent stress amplitudes, and the corresponding SN curve parameters are selected based on the bridge structure type and material parameters to obtain the equivalent stress amplitude sequence.

[0189] S38, perform hierarchical statistics on the equivalent stress amplitude sequence and calculate the fatigue damage increment.

[0190] Specifically, the equivalent stress amplitude is divided into multiple intervals, the number of cycles in each interval is counted, and the allowable number of cycles of the SN curve in the corresponding interval is combined with the linear cumulative damage criterion to calculate the fatigue damage increment in the current short-term analysis window.

[0191] S39, Update cumulative damage metrics.

[0192] Specifically, the calculated fatigue damage increment is superimposed with the historical cumulative damage value to obtain a cumulative damage index that characterizes the degree of deterioration of the bridge structure.

[0193] S310, perform trend analysis on the risk score sequence and extract statistical features.

[0194] Specifically, trend fitting analysis is performed on the risk score sequence within the short-term analysis window, and the trend slope index and trend direction indicator are adaptively determined based on the goodness of fit; at the same time, the coefficient of variation of the risk score sequence is calculated to characterize the degree of volatility.

[0195] S311 calculates the short-term trend risk score by fusing multiple indicators.

[0196] Specifically, the fatigue damage increment, trend slope index, and coefficient of variation are used as input parameters. Under the condition of satisfying the normalization constraint of the weight coefficient, a weighted fusion calculation is performed to output a short-term trend risk score.

[0197] The short-term trend risk assessment method described in this application constructs a short-term analysis window based on the traffic load triggering mechanism, and integrates fatigue damage evolution, risk change trends and fluctuation characteristics within the window to achieve dynamic characterization and quantitative assessment of the short-term risk status of bridges. This effectively reflects the risk accumulation effect and evolution trend under continuous load, providing a key basis for subsequent comprehensive risk assessment.

[0198] S4. Based on a fixed sliding time window, perform statistical analysis and modal feature extraction on the net structural response signal, calculate the structural health index, and dynamically correct the preset risk judgment threshold based on the structural health index.

[0199] The method for calculating the structural health index and dynamically adjusting the risk assessment threshold includes the following steps:

[0200] S41, construct a fixed sliding time window to segment the net structural response signal.

[0201] Specifically, the window length and sliding step size of the sliding time window are set, and the net structural response signal is continuously segmented to form a window data sequence arranged in time series.

[0202] For example, the window length can be set according to the bridge's dynamic characteristics, and the sliding step size is used to balance calculation accuracy and real-time performance.

[0203] S42 calculates statistical characteristic parameters within each sliding time window.

[0204] Specifically, within each fixed sliding time window, statistical analysis is performed on the net structural response signal to obtain statistical parameters including mean, standard deviation, peak value, peak-to-valley difference, skewness, and kurtosis, in order to characterize the distribution characteristics and fluctuation of the signal.

[0205] S43, calculate the rate of change of statistical characteristic parameters.

[0206] Specifically, based on the statistical parameters between adjacent sliding time windows, the rate of change of each statistical characteristic is calculated to reflect the dynamic trend of structural response characteristics.

[0207] S44, perform modal recognition processing to extract structural modal parameters.

[0208] Specifically, within each sliding time window, modal identification analysis is performed on the net structural response signal to extract multi-mode parameters of the structure, including natural frequency, damping ratio, and mode participation factor.

[0209] For example, the random subspace identification method or the frequency domain decomposition method can be used to extract modal parameters.

[0210] S45, screen the modal parameters and calculate the deviation index.

[0211] Specifically, the identified modal parameters are screened based on a preset stability criterion to obtain a set of stable modal parameters; and the stable modal parameters are compared and analyzed with preset benchmark values ​​to obtain the natural frequency deviation, damping ratio deviation, and modal confidence criterion values ​​for each order.

[0212] S46, construct multi-source feature input vectors.

[0213] Specifically, the statistical parameter change rate, frequency deviation, damping ratio deviation, and modal confidence criterion values ​​are collected to form a multi-source feature input vector for structural state assessment.

[0214] S47 performs membership degree transformation on multi-source features.

[0215] Specifically, membership function mapping is performed on each feature quantity in the multi-source feature input vector, and the membership value corresponding to each feature quantity is calculated based on the preset trapezoidal membership function to achieve a unified expression of features with different dimensions.

[0216] S48 is weighted and fused, and then normalized to obtain the structural health index.

[0217] Specifically, a weighted summation is performed based on the membership values ​​of each feature and a preset weight vector to obtain a comprehensive evaluation result; the comprehensive evaluation result is then normalized to output a structural health index, which is used to characterize the overall health status of the bridge.

[0218] S49, Construct a risk assessment threshold system.

[0219] Specifically, a risk assessment threshold system is established, including a first-level warning threshold, a second-level warning threshold, and a third-level warning threshold. The initial values ​​of each level of threshold are determined based on bridge design specifications and risk acceptability criteria, and are set according to a preset proportional relationship.

[0220] S410, determine the range to which the structural health index belongs and match the correction coefficient.

[0221] Specifically, the current structural health index is compared with a preset interval threshold to determine its interval, and the corresponding threshold correction coefficient group is matched according to the interval.

[0222] S411, dynamically adjusts the risk assessment threshold.

[0223] Specifically, based on the threshold correction coefficient group, the warning thresholds at each level are multiplicatively corrected: when the structural health index is in different ranges, the corresponding correction coefficients are used to adjust the warning thresholds for Level 1, Level 2 and Level 3 respectively; and the correction coefficients can be fine-tuned within a preset range according to the bridge importance level, maintenance resource constraints and traffic management strategies.

[0224] The structural health assessment and threshold dynamic correction method described in this application achieves multi-dimensional quantitative assessment of bridge structural status by integrating statistical features and modal parameters within a sliding time window, and adaptively adjusts the risk judgment threshold based on the structural health index, thereby enabling the early warning judgment results to dynamically match the actual safety margin of the structure and effectively improving the accuracy and reliability of bridge operation risk assessment.

[0225] S5 weights and integrates the single load event risk score, short-term trend risk score, and structural health index to obtain the dynamic risk index.

[0226] The method for obtaining the dynamic risk index includes the following steps:

[0227] S51 sets the basic weight coefficients for each sub-item risk indicator.

[0228] Specifically, the basic weighting coefficients for the single load event risk score, the short-term trend risk score, and the structural health index are set to 0.4, 0.4, and 0.2, respectively, to reflect the dominant role of short-term risk and single event risk, while also taking into account the overall structural health status.

[0229] S52 adaptively adjusts the weighting coefficients based on real-time data quality.

[0230] Specifically, based on the data reliability and validity of each sub-indicator, the basic weight coefficients are dynamically adjusted: when the confidence interval width of a single load event risk score exceeds a preset threshold (e.g., 15 points), its weight coefficient is reduced by a preset amount (e.g., 0.05), and the corresponding increment is evenly distributed to the short-term trend risk score and the structural health index; when the number of effective load events within the short-term analysis window is less than a preset number (e.g., 5 times), the weight coefficient of the short-term trend risk score is reduced by a preset amount, and the corresponding increment is evenly distributed to the single load event risk score; when the standard deviation of the structural health index in the most recent updates exceeds a preset threshold (e.g., 10), its weight coefficient is reduced by a preset amount, and the corresponding increment is evenly distributed to the single load event risk score and the short-term trend risk score.

[0231] S53, normalize and correct the adjusted weighting coefficients.

[0232] Specifically, the adaptively adjusted weight coefficients are normalized so that their sum satisfies the constraint of 1, while any weight coefficient is limited to a preset lower limit (such as 0.10) to avoid information loss due to excessively low weight of a single indicator.

[0233] S54 performs weighted fusion calculation to obtain the dynamic risk index.

[0234] Specifically, based on the normalized and corrected weight coefficients, a weighted linear combination of the single load event risk score, short-term trend risk score, and structural health index is performed to obtain a dynamic risk index that reflects the current comprehensive risk level of the bridge.

[0235] The dynamic risk fusion method described in this application introduces a data quality-driven adaptive weight adjustment mechanism to achieve dynamic trade-offs and optimized allocation among multi-source risk indicators, and combines normalization constraints for unified correction. This enables the stable output of highly reliable comprehensive risk assessment results under different operating conditions and data quality conditions, effectively improving the robustness and accuracy of bridge operation risk assessment.

[0236] S6 determines whether the dynamic risk index exceeds the revised risk assessment threshold and implements tiered early warning and traffic control.

[0237] The methods for determining tiered early warnings, triggering early warning signals, and implementing traffic control include the following steps:

[0238] S61, determine whether the dynamic risk index exceeds the revised risk assessment threshold.

[0239] Specifically, the currently calculated dynamic risk index is compared with the risk judgment thresholds at each level after correction by the structural health index. When the dynamic risk index does not exceed the lowest level warning threshold, the current traffic operation status is maintained, and bridge structural response data and traffic load data are continuously collected before proceeding to the next round of detection and evaluation. When the dynamic risk index exceeds any level warning threshold, the graded warning judgment process is initiated.

[0240] S62, based on the dynamic risk index, makes graded early warning judgments.

[0241] Specifically, the bridge's operational status is classified according to the relationship between the dynamic risk index and the warning thresholds at each level: when the dynamic risk index exceeds the level 3 warning threshold but does not exceed the level 2 warning threshold, it is classified as a level 3 warning, and the corresponding bridge is in a slightly abnormal state; when the dynamic risk index exceeds the level 2 warning threshold but does not exceed the level 1 warning threshold, it is classified as a level 2 warning, and the corresponding bridge is in a moderate risk state; when the dynamic risk index exceeds the level 1 warning threshold, it is classified as a level 1 warning, and the corresponding bridge is in a high risk state.

[0242] For example, during the adjustment of warning levels, if the conditions for downgrading are met, the conditions must be met for at least three consecutive assessment cycles before downgrading can be implemented; if the conditions for upgrading are met, the upgrading operation is performed directly to improve the timeliness of the warning response.

[0243] S63 triggers the corresponding level of warning signal.

[0244] Specifically, the warning signal adopts a multi-channel parallel triggering method, including the graphic display of variable message signs at bridge entrances, the activation of audible and visual alarm devices, the push of warning data packets to the traffic control center platform, and the sending of notification information to management personnel terminals. When a Level 3 warning is determined, the basic warning signal is triggered, controlling the variable message signs to display yellow warning information and pushing corresponding data. When a Level 2 warning is determined, based on the Level 3 warning, the audible and visual alarm devices are further activated and a notification is sent to management personnel, while the information signs are updated to display orange warning information. When a Level 1 warning is determined, based on the Level 2 warning, the audible and visual alarm devices are switched to continuous operation mode, the information signs are updated to display red warning information, and higher-level warning data is pushed.

[0245] S64, implement traffic control measures corresponding to the warning level.

[0246] Specifically, when a Level 3 warning is issued, basic control measures are implemented, including reducing the speed limit at the bridge entrance (the reduction should not be less than a preset percentage) and pushing the speed limit change information to the navigation service platform. When a Level 2 warning is issued, in addition to implementing Level 3 control measures, further traffic flow restriction measures are implemented. Vehicle counting devices are used to monitor traffic flow on the bridge, and when the speed exceeds a preset limit, some entrance ramp traffic lights are turned off to restrict traffic flow. When a Level 1 warning is issued, in addition to implementing Level 2 control measures, mandatory control measures are implemented, including closing the bridge entrance and activating an emergency detour guidance plan, and releasing detour route information at key points.

[0247] The aforementioned graded early warning and traffic control method of this application achieves accurate graded determination of bridge operation status by comparing dynamic risk index with adaptively corrected risk threshold. Combined with multi-channel early warning signal triggering mechanism and graded traffic control strategy, it constructs a closed-loop control system from risk identification to proactive intervention, thereby effectively improving the bridge operation safety assurance capability and traffic management response efficiency.

[0248] Figure 2 This is a schematic diagram of the architecture of a bridge traffic warning system for intelligent transportation, provided as an embodiment of the third aspect of this application.

[0249] The intelligent transportation bridge traffic warning system includes:

[0250] The data acquisition module is configured to: acquire bridge structural response data and traffic load data in real time, and perform time synchronization processing on multi-source data to generate a raw monitoring data stream with a unified timestamp;

[0251] The signal processing module, connected to the data acquisition module, is configured to: filter, denoise, and decouple the raw monitoring data stream from environmental factors, and extract the net structural response signal to eliminate the interference of environmental background and temperature effects on the structural response;

[0252] The event detection module, connected to the signal processing module, is configured to: perform load event detection and interval division based on the net structural response signal and traffic load information, and extract event features to realize the identification and classification of single load events;

[0253] The damage assessment module, connected to the event detection module, is configured to: construct a short-time analysis window for the risk score sequence of a single load event, and quantify the degree of damage accumulation and risk evolution trend of the bridge in a short period of time based on the fatigue cumulative damage model and trend analysis method.

[0254] The health monitoring module, connected to the signal processing module, is configured to: perform statistical analysis and modal feature extraction on the net structural response signal based on a sliding time window, calculate the structural health index, and comprehensively evaluate the overall state and long-term safety performance of the bridge structure.

[0255] The early warning and control module, connected to the damage assessment module and the health monitoring module, is configured to: perform a fusion calculation of the risk score of a single load event, the short-term trend risk score and the structural health index to obtain a dynamic risk index, and perform a graded early warning judgment based on the dynamic risk index and the corrected risk judgment threshold, triggering the corresponding level of early warning signal and executing the corresponding traffic control measures.

[0256] Specifically, the data acquisition module includes a structural sensor interface unit and a traffic detection interface unit; the structural sensor interface unit is used to connect to devices such as strain sensors, acceleration sensors, and displacement sensors to acquire bridge structural response data; the traffic detection interface unit is used to connect to axle load detection equipment and vehicle detection devices to acquire vehicle load and traffic information; the data acquisition module has a built-in time synchronization unit to perform unified time reference alignment processing on data with different sampling frequencies to ensure the temporal consistency of multi-source data.

[0257] Specifically, the signal processing module adopts a hierarchical processing architecture, including a preprocessing subunit and an environment decoupling subunit. The preprocessing subunit is used to filter the original signal and remove outliers. The environment decoupling subunit decomposes and reconstructs the structural response signal based on the environmental background baseline model and the temperature response model, thereby extracting the net structural response signal and improving the accuracy of subsequent analysis.

[0258] Specifically, the event detection module makes a joint judgment based on the structural response change characteristics and the axle load trigger signal, and realizes load event detection through threshold recognition and time interval division; and obtains characteristic parameters such as response peak value, duration and loading rate through the feature extraction unit to provide input for risk assessment.

[0259] Specifically, the damage assessment module includes a window construction subunit and a trend analysis subunit. The window construction subunit is used to trigger a short-term analysis window based on the statistical characteristics of load events. The trend analysis subunit calculates the damage increment and fits the trend of the risk score sequence based on the fatigue cumulative damage theory and time series analysis method, thereby obtaining a short-term trend risk score.

[0260] Specifically, the health monitoring module includes a statistical analysis subunit and a modal identification subunit. The statistical analysis subunit is used to calculate the statistical characteristic parameters of the structural response and their rate of change. The modal identification subunit is used to extract the natural frequency, damping ratio and modal participation coefficient of the structure, and calculate the structural health index based on the multi-source feature fusion method to achieve a quantitative assessment of the long-term performance of the bridge.

[0261] Specifically, the early warning and control module includes a risk fusion subunit and an early warning control subunit. The risk fusion subunit is used to weight and fuse multi-source risk indicators and generate a dynamic risk index. The early warning control subunit triggers multi-channel early warning signals based on the graded threshold judgment results and links with traffic management measures such as speed limit control, traffic flow restriction and closure control to achieve proactive intervention in bridge operation risks.

[0262] The intelligent transportation bridge traffic early warning system described in this application acquires bridge structural response data and traffic load data through a data acquisition module and performs time-synchronized processing to achieve unified management of multi-source data. A signal processing module filters and decouples the raw data from the environment to extract the net structural response signal, improving data quality. An event detection module accurately identifies and extracts features of load events, and a damage assessment module quantitatively analyzes short-term risk evolution. Simultaneously, a health monitoring module comprehensively evaluates the long-term performance of the structure and calculates the structural health index. Based on this, an early warning and control module fuses and calculates multi-source risk indicators and executes tiered early warning and traffic control, achieving closed-loop control from data acquisition and risk assessment to early warning intervention, thereby enhancing the bridge's operational safety and the level of intelligent traffic management.

[0263] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A bridge traffic warning method for intelligent transportation, characterized in that, Includes the following steps: Based on bridge structural response data and traffic load data, the net structural response signal is obtained; Load event detection and interval division are performed based on the net structural response signal, and the risk score of a single load event is calculated. Construct a short-term analysis window, perform cumulative damage calculation and trend analysis on the risk score sequence within the short-term analysis window, and obtain a short-term trend risk score; Statistical analysis and modal feature extraction of the net structural response signal are performed based on a fixed sliding time window, the structural health index is calculated, and the preset risk judgment threshold is dynamically corrected based on the structural health index. The dynamic risk index is obtained by weighting and fusing the single load event risk score, the short-term trend risk score and the structural health index. Determine whether the dynamic risk index exceeds the revised risk assessment threshold: If the risk level is exceeded, a graded early warning will be issued based on the dynamic risk index, triggering the corresponding level of early warning signal and implementing the corresponding traffic control measures. If the limit is not exceeded, the current traffic operation status will be maintained, and bridge structural response data and traffic load data will continue to be collected for the next round of testing and evaluation.

2. The bridge traffic warning method according to claim 1, characterized in that, Method for obtaining net structural response signal The bridge structural response data and traffic load data are filtered to obtain a smooth signal; Based on data from periods of low load or no vehicle traffic, an environmental background baseline for the bridge structure is obtained by fitting. The environmental background baseline is subtracted from the smoothed signal to obtain the structural response signal with the environmental background removed; Based on key section temperature data and structural response signals after removing environmental background, the temperature influence components of the bridge structure are identified. The net structural response signal is obtained by subtracting the temperature component from the structural response signal after removing the influence of the environmental background.

3. The bridge traffic warning method according to claim 2, characterized in that, The method for performing load event detection and interval division is as follows: The start and end times of the load event are determined based on the net structural response signal and the axle load information obtained by the vehicle detection equipment. The analysis interval for a single load event is constructed by using the preset duration before the start time and the preset duration after the end time as buffer boundaries. When the time interval between two adjacent load events is less than a preset interval threshold, the adjacent load events are merged into consecutive load events, and the number of merged events is recorded. When the duration of the continuous load event exceeds a preset duration threshold, the continuous load event is segmented based on the local minimum point of the net structural response signal, and the duration of each segment is not less than the preset minimum duration.

4. The bridge traffic warning method according to claim 3, characterized in that, The method for calculating the risk score of a single load event is as follows: Based on the defined load event analysis interval, load event characteristic quantities are extracted, including the net structural response peak value, the ratio of the peak value to the bridge design bearing capacity response benchmark value, the response rise slope, the response duration, the response integral area, the interval between adjacent load events, the estimated total axle load of vehicles passing at the same time, and the multi-lane synchronous loading marker. The load event features are input into a pre-trained gradient boosting regression model to perform risk mapping calculations and obtain a risk score for a single load event.

5. The bridge traffic warning method according to claim 1, characterized in that, The method for constructing a short-term analysis window is as follows: The frequency of load events is statistically analyzed based on a fixed time step, and the statistical features within each time step are obtained and stored in a counting buffer. The statistical features include: the total number of load events, the number of high-risk events, and the number of simultaneous over-limit events in both directions of the lane. Based on the cumulative statistical results within three consecutive fixed time steps, a condition judgment is made. When the total number of cumulative load events is not less than the preset number threshold, or the number of high-risk events is not less than the preset risk number threshold, or the number of bidirectional synchronization over-limit events is not less than the preset synchronization number threshold, the construction of a short-term analysis window is triggered. When the construction conditions are triggered, the current time is used as the end point of the window, and three fixed time steps are traced back as the starting point of the window to form a short-term analysis window.

6. The bridge traffic warning method according to claim 1, characterized in that, The method for obtaining the short-term trend risk score is as follows: Equivalent stress mapping is performed on the risk scores of single load events arranged in time series within the short-time analysis window, and the corresponding SN curve parameters are selected based on the bridge structure type and material parameters to obtain the equivalent stress amplitude sequence. The equivalent stress amplitude sequence is divided into hierarchical intervals to construct stress amplitude distribution intervals, and the number of cycles within each distribution interval is counted. Based on the number of cycles in each stress amplitude range and the allowable number of cycles for the SN curve in the corresponding range, the fatigue damage increment in the current short-time analysis window is calculated according to the linear cumulative damage criterion. The fatigue damage increment is superimposed with the historical cumulative damage value to obtain the cumulative damage index characterizing the degree of structural deterioration. Trend fitting analysis is performed on the risk scores of single load events arranged in time series within the short-term analysis window, and the trend slope index and trend direction indicator are adaptively determined based on the goodness of fit. The coefficient of variation is calculated based on the risk score sequence of a single load event within a short-time analysis window. The fatigue damage increment, trend slope index, and coefficient of variation are used as input parameters for fusion calculation. Under the condition of satisfying the normalization constraint of weight coefficient, the short-term trend risk score is obtained.

7. The bridge traffic warning method according to claim 1, characterized in that, The method for statistical analysis and modal feature extraction of net structural response signals based on a fixed sliding time window is as follows: A fixed sliding time window is constructed, and the window length and sliding step size are set. The net structural response signal is segmented to obtain a window data sequence arranged in time series. Within each fixed sliding time window, statistical characteristics of the net structural response signal are calculated to obtain statistical parameters including mean, standard deviation, peak value, peak-to-valley difference, skewness, and kurtosis. Calculate the rate of change of statistical parameters based on adjacent sliding time windows; Within each sliding time window, modal identification processing is performed on the net structural response signal to extract multi-mode parameters of the structure, including natural frequency, damping ratio and mode participation coefficient. Based on a preset stability criterion, modal parameters are screened to obtain a set of stable modal parameters; By comparing and analyzing the stable modal parameters with the preset deviation criteria, the natural frequency deviation, damping ratio deviation, and modal confidence criterion values ​​for each order are obtained.

8. The bridge traffic warning method according to claim 7, characterized in that, The method for calculating the structural health index is as follows: The rate of change of statistical parameters, the frequency deviations of each order, the damping ratio deviations, and the modal confidence criterion values ​​are collected to obtain a multi-source feature input vector; The membership degree transformation process is performed on each feature quantity in the multi-source feature input vector, and the membership degree value corresponding to each feature quantity is obtained based on the preset trapezoidal membership degree function. The comprehensive evaluation result is obtained by performing a weighted summation based on the membership values ​​corresponding to each feature and the preset weight vector. The comprehensive evaluation results are normalized to obtain the structural health index.

9. The bridge traffic warning method according to claim 1, characterized in that, The dynamic risk index is obtained as follows: The basic weighting coefficients for the single load event risk score, short-term trend risk score, and structural health index are set to 0.4, 0.4, and 0.2, respectively. The basic weighting coefficients are adaptively adjusted based on real-time data quality: when the confidence interval width of a single load event risk score exceeds 15 points, its weighting coefficient is reduced by 0.05, and the corresponding increment is evenly distributed to the short-term trend risk score and the structural health index; when the number of effective load events within the short-term analysis window is less than 5, the weighting coefficient of the short-term trend risk score is reduced by 0.05, and the corresponding increment is evenly distributed to the single load event risk score; when the standard deviation of the structural health index's three most recent update values ​​exceeds 10, its weighting coefficient is reduced by 0.05, and the corresponding increment is evenly distributed to the single load event risk score and the short-term trend risk score. The adaptively adjusted weight coefficients are corrected so that the sum of the weight coefficients satisfies the normalization constraint and that no weight coefficient is lower than 0.

10. Based on the corrected weighting coefficients, the single load event risk score, short-term trend risk score, and structural health index are weighted and linearly combined to obtain the dynamic risk index.

10. A bridge traffic warning system for intelligent transportation, used to implement the bridge traffic warning method as described in any one of claims 1-9, comprising: The data acquisition module is used to collect bridge structural response data and traffic load data in real time. The signal processing module is used to acquire the net structural response signal of the bridge structure; The event detection module is used to realize the real-time identification and classification of load events; The damage assessment module is used to quantify the degree of cumulative damage and development trend of bridges in a short period of time. The health monitoring module is used to comprehensively assess the overall condition and long-term safety performance of the bridge structure. The early warning and control module is responsible for triggering early warning signals of corresponding levels to achieve proactive management of bridge safety operation.

Images