Bridge load intelligent management method and system based on artificial intelligence, and medium

Abstract

The invention discloses a bridge load intelligent management method and system based on artificial intelligence and a medium, and belongs to the technical field of crossing of bridge engineering and artificial intelligence. The method comprises the following steps: S1, acquiring vehicle information, bridge information and environment information by using various monitoring devices to obtain first bridge data; s2, performing data preprocessing on the first bridge data to obtain second bridge data; s3, constructing a prediction model; identifying potential risks in advance to obtain an analysis result; s4, automatically adjusting a traffic flow control strategy of the bridge according to an analysis result; s5, continuously monitoring the change condition of the bridge state, and updating the prediction model and the traffic flow control strategy according to the change condition; and S6, recording the result of adjusting the traffic flow control strategy each time and feedback information, and continuously iterating the prediction model according to the feedback information. A sensor network and an artificial intelligence technology are utilized, real-time monitoring, prediction and optimization adjustment of the bridge load are achieved, and it is ensured that the bridge structure is in a safe state all the time.

Description

Technical Field

[0001] This invention relates to the field of bridge engineering and artificial intelligence, and in particular to an intelligent bridge load management method, system and medium based on artificial intelligence. Background Technology

[0002] Currently, with the acceleration of urbanization and the increase in traffic flow, the structural safety and load management of bridges, as an important component of transportation infrastructure, are receiving increasing attention. Traditional bridge load monitoring methods mainly rely on periodic manual inspections and single-sensor monitoring, which suffer from problems such as data update lag, poor real-time performance, and limited predictive capabilities, making it difficult to detect safety hazards such as overloading and fatigue damage in a timely manner. In addition, traditional methods often adopt static management strategies, lacking the ability to dynamically optimize and control bridge loads, making it difficult to adapt to complex traffic environments and changes in bridge structural conditions. This results in bridges being under high loads or uneven stress for extended periods, accelerating structural aging and damage. Summary of the Invention

[0003] The purpose of this invention is to overcome the shortcomings of the prior art and provide a bridge load intelligent management method, system and medium based on artificial intelligence.

[0004] The objective of this invention is achieved through the following technical solution: The first aspect of this invention provides: an intelligent bridge load management method based on artificial intelligence, comprising the following steps: S1: In the data acquisition phase, various monitoring devices are used to collect vehicle information, bridge information, and environmental information to obtain the first bridge data; S2: Data preprocessing stage, the first bridge data is preprocessed to obtain the second bridge data; S3: In the model training phase, the artificial intelligence algorithm model is trained using the second bridge data to obtain the prediction model; the prediction model is used to analyze real-time bridge data to predict the impact of dynamic bridge loads on the bridge structure and identify potential risks in advance to obtain analysis results. S4: Management optimization phase, automatically adjust the bridge's traffic flow control strategy based on the analysis results; S5: Continuous monitoring phase, continuously monitor changes in bridge status, and update prediction models and traffic flow control strategies based on changes; S6: Feedback Iteration Phase. Record the results and feedback information of each adjustment to the traffic flow control strategy, and continuously iterate the prediction model based on the feedback information.

[0005] Preferably, the S1: data acquisition stage further includes the following steps: The vehicle's location information is obtained in real time through the GPS system, and the vehicle's weight information is collected through the WIM dynamic weighing system installed on the bridge. The vehicle's speed is calculated based on the location information and time difference. Let the vehicle's position at time t1 be (x1, y1) and its position at time t2 be (x2, y2). Calculate the distance d between the two points using the following formula: ; The vehicle's speed v is calculated using the following formula: ,in For time difference; The strain gauge measures the deformation of the bridge under load. Let the original length of the bridge be L0, and the length after deformation under load be L. Then the strain... for: ; The wind speed measured using an anemometer is V. w The wind force F is calculated using the following formula. w : Where ρ is the air density, A is the windward area, and C is the air density. d This is the drag coefficient.

[0006] Preferably, the S2: data preprocessing stage further includes the following steps: Data cleaning is performed by identifying and removing outliers caused by equipment failure, transmission errors, or human interference using the Z-score method. Data standardization is achieved through the min-max normalization method, enabling data from different sources and types to be compared and analyzed on the same scale. When the absolute value of a data point's Z-score exceeds a set threshold, the data point is considered an outlier. The formula for calculating the Z-score is: , where x is the value of the data point, μ is the mean of the sample set, and σ is the standard deviation of the sample set; if |Z| of a data point > 3, then the data point is an outlier and should be removed. The formula for data standardization using the min-max normalization method is as follows: , where X norm The dataset is the normalized dataset, where X is the original data. min X is the minimum value in the dataset. max This represents the maximum value in the dataset.

[0007] Preferably, the S3: model training phase further includes the following steps: The prediction model identifies key factors affecting bridge load and predicts bridge load response or bridge health status. During training, the accuracy of the prediction model is improved through loss functions and optimization algorithms, and cross-validation is used to enhance the generalization ability of the prediction model. The cleaned, formatted, and standardized dataset The data points are input into the prediction model, where each data point represents an observation sample. Through correlation analysis or internal model weight evaluation, the features with the greatest impact on bridge load prediction are selected. For dataset X... norm Assuming p features are selected, the input feature matrix is: ,in This represents the p-th feature of the n-th observation sample; When training a prediction model, a neural network is selected to perform a nonlinear mapping from the input features X. input To the predicted value Training, predicted values Generate based on the prediction target of bridge load impact: ,in It is the activation function of the neural network; θ are the model parameters, representing the weights and biases of the neural network; The model parameters are optimized by minimizing the loss function. If the training objective is to minimize the mean squared error, then the loss function is: ,in y is the predicted bridge load value for the i-th sample predicted by the prediction model. i is the actual value of the bridge load for the i-th sample, and N is the number of samples; We optimize using the gradient descent algorithm, updating the model parameters θ to minimize the loss function. Where η is the learning rate, It is the gradient of the loss function with respect to the model parameters θ; The test is performed using a validation set, and the final evaluation result is the average of all validation results. ,in It is the loss function of the i-th validation set; Regularization is used to limit the complexity of the prediction model and prevent overfitting. The L2 regularization loss function is: , where λ is the regularization coefficient and m is the number of model parameters θ.

[0008] Preferably, the S4: management optimization phase further includes the following steps: By combining and analyzing real-time monitoring data and model prediction results, traffic flow control strategies are adjusted to optimize bridge load distribution, reduce uneven bridge loads, and avoid bridge overloading. Based on the load response predicted by the model, the response state of the bridge structure under different load conditions is analyzed. Let the response of the bridge be y. bridge The prediction model then uses the vehicle load data x as input. load and environmental condition data x env Predicting the bridge's response: ,in This represents the mapping of the load output by the prediction model to the bridge structure; Using the bridge response results output by the predictive model, optimize traffic flow and vehicle speed, and set the objective function. To minimize the maximum stress or deformation of the bridge and avoid local overload, an objective function is set. A comprehensive evaluation standard for the load response of bridges: ,in It is the predicted response value of the i-th bridge monitoring point. This is the current traffic flow configuration; Based on the objective function Generate a traffic flow control strategy to minimize the load on the bridge by optimizing traffic flow and vehicle speed. Define the optimization problem as follows: .

[0009] Preferably, the S5: continuous monitoring phase further includes the following steps: The bridge's health status data is collected in real time using sensors, and the bridge status data is set as follows: Where m1 represents the number of sensors monitored in real time, y m1 This represents the real-time observation value of the m1-th sensor; Will The input is fed into the prediction model for real-time prediction. Let the output of the prediction model be... The predicted load response of the bridge under its current condition is calculated using the following formula: ,in For the prediction function of the prediction model; When the bridge's monitoring points detect overload, the system automatically adjusts traffic flow or speed limits to reduce bridge load. Let the bridge's maximum load response be... The calculation formula is: The traffic flow control strategy is dynamically updated by adjusting the current traffic flow configuration: ,in For the new traffic flow control strategy The adjusted traffic flow configuration is represented by α, where α is the adjustment coefficient. For the predicted maximum bridge load response, This represents the actual monitored maximum bridge load response.

[0010] Preferably, the S6: feedback iteration stage further includes the following steps: The adjusted traffic flow control strategy, bridge load response data, bridge health status before and after the adjustment, and traffic flow conditions are recorded and represented as follows: ,in This represents the real-time data of the nth sensor after the t-th adjustment, where n is the number of sensors detected in the adjustment strategy; After recording the results of each adjustment, the bridge load, stress, and location parameters before and after the adjustment were compared to assess whether the adjusted traffic flow control strategy effectively reduced the bridge load and to analyze whether the adjustment measures had a positive impact on bridge health. Error analysis was used to quantitatively evaluate the adjustment effect, comparing predicted values ​​with actual values ​​and calculating the mean absolute error. for: ,in It is a predicted value; These are actual observed values; Mean absolute error, representing the prediction accuracy of each adjustment measure; Based on the feedback information of the adjustment effect, the parameters of the prediction model are improved through incremental learning, and the prediction model is optimized through gradient descent algorithm during the update. Based on the updated prediction model, combined with the mean absolute error Further optimize traffic flow control strategies; After each round of feedback and optimization, new predictive models and traffic flow control strategies will be iterated.

[0011] A second aspect of the present invention provides: an intelligent bridge load management system based on artificial intelligence, used to implement any of the above-mentioned intelligent bridge load management methods based on artificial intelligence, comprising: The data acquisition module is used to collect vehicle information, bridge information, and environmental information using various monitoring devices to obtain the first bridge data; The data preprocessing module is used to preprocess the first bridge data to obtain the second bridge data; The model training module is used to train an artificial intelligence algorithm model using the second bridge data to obtain a prediction model; by analyzing real-time bridge data through the prediction model, the impact of dynamic bridge loads on the bridge structure is predicted, and potential risks are identified in advance to obtain analysis results. The management optimization module is used to automatically adjust the traffic flow control strategy for bridges based on the analysis results; The continuous monitoring module is used to continuously monitor changes in the bridge's condition and update the prediction model and traffic flow control strategy based on these changes. The feedback iteration module records the results and feedback information of each adjustment to the traffic flow control strategy, and continuously iterates the prediction model based on the feedback information.

[0012] A third aspect of the present invention provides: a computer-readable storage medium storing computer-executable instructions, wherein when the computer-executable instructions are loaded and executed by a processor, any of the above-mentioned intelligent bridge load management methods based on artificial intelligence are implemented.

[0013] The beneficial effects of this invention are: 1) Real-time monitoring and dynamic adjustment: Traditional bridge load management methods typically rely on periodic inspections, which are subject to delays. This invention utilizes sensor networks and artificial intelligence technology to achieve real-time monitoring, prediction, and optimized adjustment of bridge loads, ensuring that the bridge structure is always in a safe state.

[0014] 2) Intelligent prediction for enhanced safety: This invention employs a deep learning model to analyze historical load data and real-time monitoring data, predicting future load trends for the bridge and enabling proactive preventative measures to avoid sudden structural damage. Compared to traditional statistical analysis methods, it offers higher prediction accuracy and faster response time.

[0015] 3) Optimize load distribution and extend bridge life: Based on artificial intelligence, calculate the stress distribution and fatigue accumulation state of the bridge, optimize traffic flow and vehicle passage strategies, such as dynamic flow restriction, intelligent speed limit, lane adjustment, etc., thereby reducing local stress concentration and delaying bridge aging.

[0016] 4) Adaptive feedback mechanism to improve management efficiency: This invention introduces reinforcement learning and incremental learning techniques, enabling the system to continuously optimize its own model based on historical data, adapt to different environments and load conditions, reduce manual intervention, and improve management efficiency.

[0017] 5) Integrating multiple monitoring methods to improve data accuracy: This invention combines multiple technologies such as strain sensors, accelerometers, displacement sensors, and fiber optic grating monitoring, and integrates multi-source data to improve the accuracy and reliability of load assessment, which is more advantageous than a single monitoring method.

[0018] 6) Reduce maintenance costs and improve operational efficiency: Through intelligent management, it is possible to provide early warnings and adjust operational strategies before the bridge load reaches a critical state, avoiding premature or excessive repairs, thereby reducing maintenance costs and improving the bridge's operational efficiency and service life. Attached Figure Description

[0019] Figure 1This is a flowchart of the method of the present invention; Figure 2 Here is a flowchart of the data preprocessing process; Figure 3 This is a schematic diagram of the training loss curve; Figure 4 To optimize the before-and-after comparison images; Figure 5 A schematic diagram for optimizing real-time monitoring and feedback. Detailed Implementation

[0020] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] See Figures 1-5 The first aspect of this invention provides: an intelligent bridge load management method based on artificial intelligence, comprising the following steps: S1: In the data acquisition phase, various monitoring devices are used to collect vehicle information, bridge information, and environmental information to obtain the first bridge data; S2: Data preprocessing stage, the first bridge data is preprocessed to obtain the second bridge data; S3: In the model training phase, the artificial intelligence algorithm model is trained using the second bridge data to obtain the prediction model; the prediction model is used to analyze real-time bridge data to predict the impact of dynamic bridge loads on the bridge structure and identify potential risks in advance to obtain analysis results. S4: Management optimization phase, automatically adjust the bridge's traffic flow control strategy based on the analysis results; S5: Continuous monitoring phase, continuously monitor changes in bridge status, and update prediction models and traffic flow control strategies based on changes; S6: Feedback Iteration Phase. Record the results and feedback information of each adjustment to the traffic flow control strategy, and continuously iterate the prediction model based on the feedback information.

[0022] In this embodiment, precise monitoring, intelligent prediction, and dynamic optimization of bridge loads are achieved through multi-source sensor data fusion, intelligent algorithm modeling, and real-time prediction and optimization control. Specifically, this method utilizes sensing devices such as GPS, dynamic weighing systems, strain gauges, and anemometers to collect real-time data on the bridge and its traffic environment, and improves data quality through preprocessing methods such as data cleaning and standardization. Subsequently, the stress state of the bridge is predicted based on artificial intelligence algorithms (such as neural networks), and the model is optimized using incremental learning methods to enable it to adapt to changes in bridge loads. Through real-time analysis and optimization control, this method can automatically adjust traffic flow, optimize load distribution, and provide a feedback mechanism to continuously improve management strategies, effectively overcoming the limitations of traditional bridge load management methods and improving bridge structural safety and operational efficiency.

[0023] In some embodiments, the S1: data acquisition phase further includes the following steps: The vehicle's location information is obtained in real time through the GPS system, and the vehicle's weight information is collected through the WIM dynamic weighing system installed on the bridge. The vehicle's speed is calculated based on the location information and time difference. Let the vehicle's position at time t1 be (x1, y1) and its position at time t2 be (x2, y2). Calculate the distance d between the two points using the following formula: ; The vehicle's speed v is calculated using the following formula: ,in For time difference; The strain gauge measures the deformation of the bridge under load. Let the original length of the bridge be L0, and the length after deformation under load be L. Then the strain... for: ; The wind speed measured using an anemometer is V. w The wind force F is calculated using the following formula. w : Where ρ is the air density, A is the windward area, and C is the air density. d This is the drag coefficient.

[0024] In some embodiments, the S2: data preprocessing stage further includes the following steps: Data cleaning is performed by identifying and removing outliers caused by equipment failure, transmission errors, or human interference using the Z-score method. Data standardization is achieved through the min-max normalization method, enabling data from different sources and types to be compared and analyzed on the same scale. When the absolute value of a data point's Z-score exceeds a set threshold, the data point is considered an outlier. The formula for calculating the Z-score is: , where x is the value of the data point, μ is the mean of the sample set, and σ is the standard deviation of the sample set; if |Z| of a data point > 3, then the data point is an outlier and should be removed. The formula for data standardization using the min-max normalization method is as follows: , where X norm The dataset is the normalized dataset, where X is the original data. min X is the minimum value in the dataset. max This represents the maximum value in the dataset.

[0025] In this embodiment, the threshold is set to 3 or -3. In some embodiments, the S3: model training phase further includes the following steps: The prediction model identifies key factors affecting bridge load and predicts bridge load response or bridge health status. During training, the accuracy of the prediction model is improved through loss functions and optimization algorithms, and cross-validation is used to enhance the generalization ability of the prediction model. The cleaned, formatted, and standardized dataset The data points are input into the prediction model, where each data point represents an observation sample. Through correlation analysis or internal model weight evaluation, the features with the greatest impact on bridge load prediction are selected. For dataset X... norm Assuming p features are selected, the input feature matrix is: ,in This represents the p-th feature of the n-th observation sample; When training a prediction model, a neural network is selected to perform a nonlinear mapping from the input features X. input To the predicted value Training, predicted values Generate based on the prediction target of bridge load impact: ,in It is the activation function of the neural network; θ are the model parameters, representing the weights and biases of the neural network; The model parameters are optimized by minimizing the loss function. If the training objective is to minimize the mean squared error, then the loss function is: ,in y is the predicted bridge load value for the i-th sample predicted by the prediction model. i is the actual value of the bridge load for the i-th sample, and N is the number of samples; We optimize using the gradient descent algorithm, updating the model parameters θ to minimize the loss function. Where η is the learning rate, It is the gradient of the loss function with respect to the model parameters θ; The test is performed using a validation set, and the final evaluation result is the average of all validation results. ,in It is the loss function of the i-th validation set; Regularization is used to limit the complexity of the prediction model and prevent overfitting. The L2 regularization loss function is: , where λ is the regularization coefficient and m is the number of model parameters θ.

[0026] In this embodiment, the accuracy of predictions is improved and the level of intelligence in bridge load management is enhanced through loss functions and optimization algorithms. Cross-validation is used to ensure the model's generalization ability on different datasets, thereby improving the accuracy and reliability of prediction results.

[0027] In some embodiments, the S4: management optimization phase further includes the following steps: By combining and analyzing real-time monitoring data and model prediction results, traffic flow control strategies are adjusted to optimize bridge load distribution, reduce uneven bridge loads, and avoid bridge overloading. Based on the load response predicted by the model, the response state of the bridge structure under different load conditions is analyzed. Let the response of the bridge be y. bridge The prediction model then uses the vehicle load data x as input. load and environmental condition data x env Predicting the bridge's response: ,in This represents the mapping of the load output by the prediction model to the bridge structure; Using the bridge response results output by the predictive model, optimize traffic flow and vehicle speed, and set the objective function. To minimize the maximum stress or deformation of the bridge and avoid local overload, an objective function is set. A comprehensive evaluation standard for the load response of bridges: ,in It is the predicted response value of the i-th bridge monitoring point. This is the current traffic flow configuration; Based on the objective function Generate a traffic flow control strategy to minimize the load on the bridge by optimizing traffic flow and vehicle speed. Define the optimization problem as follows: .

[0028] In this embodiment, traffic flow configuration includes speed limits, lane allocation, etc.

[0029] In some embodiments, the S5: continuous monitoring phase further includes the following steps: The bridge's health status data is collected in real time using sensors, and the bridge status data is set as follows: Where m1 represents the number of sensors monitored in real time, y m1 This represents the real-time observation value of the m1-th sensor; Will The input is fed into the prediction model for real-time prediction. Let the output of the prediction model be... The predicted load response of the bridge under its current condition is calculated using the following formula: ,in For the prediction function of the prediction model; When the bridge's monitoring points detect overload, the system automatically adjusts traffic flow or speed limits to reduce bridge load. Let the bridge's maximum load response be... The calculation formula is: The traffic flow control strategy is dynamically updated by adjusting the current traffic flow configuration: ,in For the new traffic flow control strategy The adjusted traffic flow configuration is represented by α, where α is the adjustment coefficient. For the predicted maximum bridge load response, This represents the actual monitored maximum bridge load response.

[0030] In this embodiment, the latest bridge health data is dynamically acquired and input into the artificial intelligence model to adjust the bridge load management strategy. This step ensures that the system can respond to changes in the bridge in real time, and by updating the analysis model and control strategy, the system can adapt to different bridge management needs, ensuring that bridge load management is always in an optimal state. Over time, the health status of the bridge and the external environment may change, leading to a decrease in the model's prediction accuracy. The system automatically updates the model to adapt to changes in the bridge's condition by monitoring data and prediction errors in real time. New bridge condition data can be used to train an incremental learning model to avoid model performance degradation.

[0031] In some embodiments, S6: the feedback iteration phase further includes the following steps: The adjusted traffic flow control strategy, bridge load response data, bridge health status before and after the adjustment, and traffic flow conditions are recorded and represented as follows: ,in This represents the real-time data of the nth sensor after the t-th adjustment, where n is the number of sensors detected in the adjustment strategy; After recording the results of each adjustment, the bridge load, stress, and location parameters before and after the adjustment were compared to assess whether the adjusted traffic flow control strategy effectively reduced the bridge load and to analyze whether the adjustment measures had a positive impact on bridge health. Error analysis was used to quantitatively evaluate the adjustment effect, comparing predicted values ​​with actual values ​​and calculating the mean absolute error. for: ,in It is a predicted value; These are actual observed values; Mean absolute error, representing the prediction accuracy of each adjustment measure; Based on the feedback information of the adjustment effect, the parameters of the prediction model are improved through incremental learning, and the prediction model is optimized through gradient descent algorithm during the update. Based on the updated prediction model, combined with the mean absolute error Further optimize traffic flow control strategies; After each round of feedback and optimization, new predictive models and traffic flow control strategies will be iterated.

[0032] In this embodiment, by recording the results and feedback information of each adjustment measure, the artificial intelligence algorithm model is continuously optimized, thereby improving the system's accuracy, efficiency, and responsiveness. This step ensures that the bridge load management system can be continuously adjusted and improved in practical applications to adapt to changes or unexpected situations that may occur during bridge load management. Based on the feedback information of the adjustment effects, the artificial intelligence model is optimized. By analyzing the errors and effects of each adjustment, the system can gradually optimize the model's predictive ability and improve its parameters. This process is typically carried out using incremental learning methods. The goal of incremental learning is to update the model based on new feedback data without retraining the entire model.

[0033] Based on the updated model results, the control strategy will be further optimized according to the new model. For example, after each adjustment, if the system detects that certain control measures (such as speed limits and traffic flow adjustments) are insufficient to effectively control the bridge load, the system will automatically adjust the control strategy through optimization algorithms. The goal of optimizing the control strategy is to minimize bridge load fluctuations and maintain the safety of the bridge structure.

[0034] After each round of feedback and optimization, the system continues with the next round of real-time monitoring and adjustment based on the latest model and control strategy. As the feedback and optimization cycle continues, the system can gradually improve the model's prediction accuracy and optimize the control strategy, enabling the bridge load management system to better adapt to different bridge operating conditions and traffic patterns.

[0035] The following is an example: A large highway bridge in a certain city experiences a daily traffic flow of over 100,000 vehicles, with heavy trucks accounting for approximately 15%. Built 20 years ago, the bridge has suffered varying degrees of fatigue damage and localized deformation in recent years due to increased traffic volume and frequent heavy vehicle traffic. Traditional bridge load management methods rely primarily on periodic manual inspections and single-sensor monitoring, which suffer from problems such as delayed data updates, poor real-time performance, and limited predictive capabilities, making it difficult to promptly detect safety hazards such as overloading and fatigue damage.

[0036] To ensure the structural safety of bridges and avoid potential risks caused by overloading and uneven loads, bridge management authorities decided to introduce an intelligent bridge load management system. The artificial intelligence-based intelligent bridge load management method described in this patent can be applied to this scenario, enabling real-time monitoring, prediction, and optimized management of bridge loads.

[0037] 1. Bridge Data Acquisition: GPS devices, a dynamic weighing system (WIM), strain gauges, and anemometers are installed at key locations on the bridge. GPS devices collect real-time vehicle location information, the WIM system collects vehicle weight information, strain gauges measure the degree of bridge deformation, and anemometers measure wind speed and other environmental information. Sample data is shown in Table 1. Table 1 A truck is located at (x1, y1) at time t1 and at (x2, y2) at time t2. Calculate the truck's speed v using the formula: The dynamic weighing system measures the vehicle's pressure F, and calculates the vehicle's weight W using the sensitivity coefficient S. Strain gauges measure the deformation of a bridge under load. The original length of the bridge is L0, and the length after deformation under load is L. The strain ϵ can then be expressed as: The anemometer measures the wind speed as V. w The wind force F is calculated using the following formula. w : Where ρ is the air density, A is the windward area, and C is the air density. d This is the drag coefficient.

[0038] 2. Data preprocessing: such as Figure 2 As shown, the Z-score method is used to detect and remove outliers. For example, in a collection of vehicle weight data, the Z-score of a certain vehicle's weight W is 4.5, exceeding the threshold of 3, and is therefore identified as an outlier and removed. The min-max normalization method is used to standardize the data, ensuring that data from different sources are compared on the same scale. For example, after standardization, the Z-score of the vehicle weight data W is X... norm Represented as: During the data cleaning process, outliers caused by equipment failure, transmission errors, or human interference are removed to ensure data quality and availability.

[0039] 3. Training the model: The training loss curve is as follows Figure 3 As shown, the preprocessed data is input into a neural network model to train the model for predicting bridge load response. The input feature matrix X... input Features include vehicle weight, speed, and bridge strain. ; through loss function The gradient descent algorithm optimizes the model parameters θ and minimizes the prediction error. Cross-validation is used to ensure the model's generalization ability on different datasets, thereby improving the accuracy and reliability of prediction results.

[0040] 4. Analysis and Optimization: Based on the bridge load response y predicted by the model... bridge The model analyzes the structural response of bridges under different load conditions. For example, the model predicts the maximum stress on the bridge at a certain moment. The system automatically adjusts traffic flow control strategies to optimize bridge load distribution. A target function is set. To minimize the maximum stress on the bridge: By optimizing traffic flow and vehicle speed, the load pressure on the bridge is minimized. Traffic flow control measures such as speed limits and lane allocation are adjusted. Table 2 shows some of the bridge load management effects before and after optimization. Table 2 5. Real-time monitoring and updating: Real-time collection of bridge health status data. real-time This data is then input into an artificial intelligence model for real-time prediction. For example, the real-time monitoring data of a bridge at a certain moment is... The model outputs the predicted bridge load response. ; Figure 4 It demonstrates the actual effects of bridge load management and compares data before and after optimization.

[0041] Based on real-time prediction results, traffic flow control strategies are dynamically adjusted. For example, when a bridge monitoring point detects overload, the system automatically adjusts traffic flow or speed limits to reduce the bridge load. Over time, the health status of the bridge and the external environment may change, leading to a decrease in the model's prediction accuracy. The system automatically updates the model to adapt to changes in bridge condition by monitoring data and prediction errors in real time.

[0042] 6. Feedback and Optimization: Record the results of each adjustment measure. adjust And analyze the adjustment effect. For example, the bridge load response data after a certain adjustment is... Calculation error Based on the feedback, optimize the artificial intelligence model. For example, update the model parameters θ using incremental learning methods: The control strategy is optimized based on the updated model. For example, after each adjustment, if the system detects that certain control measures are insufficient to effectively control the bridge load, the system will automatically adjust the control strategy through optimization algorithms. Figure 5 The system's dynamic adjustment process was demonstrated.

[0043] A second aspect of the present invention provides: an intelligent bridge load management system based on artificial intelligence, used to implement any of the above-mentioned intelligent bridge load management methods based on artificial intelligence, comprising: The data acquisition module is used to collect vehicle information, bridge information, and environmental information using various monitoring devices to obtain the first bridge data; The data preprocessing module is used to preprocess the first bridge data to obtain the second bridge data; The model training module is used to train an artificial intelligence algorithm model using the second bridge data to obtain a prediction model; by analyzing real-time bridge data through the prediction model, the impact of dynamic bridge loads on the bridge structure is predicted, and potential risks are identified in advance to obtain analysis results. The management optimization module is used to automatically adjust the traffic flow control strategy for bridges based on the analysis results; The continuous monitoring module is used to continuously monitor changes in the bridge's condition and update the prediction model and traffic flow control strategy based on these changes. The feedback iteration module records the results and feedback information of each adjustment to the traffic flow control strategy, and continuously iterates the prediction model based on the feedback information.

[0044] A third aspect of the present invention provides: a computer-readable storage medium storing computer-executable instructions, wherein when the computer-executable instructions are loaded and executed by a processor, any of the above-mentioned intelligent bridge load management methods based on artificial intelligence are implemented.

[0045] This invention achieves real-time monitoring, prediction, and optimized management of bridge loads through intelligent means, ensuring the safety and stability of bridge structures. The method includes the following key steps: data acquisition, utilizing sensors such as GPS, dynamic weighing systems, strain gauges, and anemometers to collect real-time information on vehicles, bridges, and the environment, ensuring data comprehensiveness and accuracy; data preprocessing, employing the Z-score method to detect outliers and standardizing data using the min-max normalization method to improve data quality and usability; model training, based on artificial intelligence algorithms (such as neural networks), predicting the impact of dynamic vehicle loads on bridge structures and optimizing load management strategies; analysis and optimization, combining AI prediction results with real-time monitoring data to automatically adjust traffic flow control measures and optimize bridge load distribution; real-time monitoring and updating, using incremental learning to dynamically adjust model parameters to adapt to changes in bridge condition; and feedback optimization, recording the results of adjustment measures and continuously optimizing the artificial intelligence algorithm model to improve the system's accuracy and efficiency. This invention innovatively combines multi-source sensor data, machine learning modeling, and incremental learning optimization strategies, overcoming the problems of strong lag and poor adaptability in traditional bridge load management, and achieving intelligent and dynamic optimized bridge load management.

[0046] The above description is merely a preferred embodiment of the present invention. It should be understood that the present invention is not limited to the forms disclosed herein and should not be construed as excluding other embodiments. It can be used in various other combinations, modifications, and environments, and can be altered within the scope of the concept described herein through the above teachings or related technologies or knowledge. Modifications and variations made by those skilled in the art that do not depart from the spirit and scope of the present invention should be within the protection scope of the appended claims.

Claims

1. A bridge load intelligent management method based on artificial intelligence, characterized in that: Includes the following steps: S1: In the data acquisition phase, various monitoring devices are used to collect vehicle information, bridge information, and environmental information to obtain the first bridge data; S2: Data preprocessing stage, the first bridge data is preprocessed to obtain the second bridge data; S3: In the model training phase, the artificial intelligence algorithm model is trained using the second bridge data to obtain the prediction model; the prediction model is used to analyze real-time bridge data to predict the impact of dynamic bridge loads on the bridge structure and identify potential risks in advance to obtain analysis results. S4: Management optimization phase, automatically adjust the bridge's traffic flow control strategy based on the analysis results; S5: Continuous monitoring phase, continuously monitor changes in bridge status, and update prediction models and traffic flow control strategies based on changes; S6: Feedback Iteration Phase. Record the results and feedback information of each adjustment to the traffic flow control strategy, and continuously iterate the prediction model based on the feedback information.

2. The bridge load intelligent management method based on artificial intelligence according to claim 1, characterized in that: The S1: data acquisition phase further includes the following steps: The vehicle's location information is obtained in real time through the GPS system, and the vehicle's weight information is collected through the WIM dynamic weighing system installed on the bridge. The vehicle's speed is calculated based on the location information and time difference. Let the vehicle's position at time t1 be (x1, y1) and its position at time t2 be (x2, y2). Calculate the distance d between the two points using the following formula: ; The vehicle's speed v is calculated using the following formula: ,in For time difference; The strain gauge measures the deformation of the bridge under load. Let the original length of the bridge be L0, and the length after deformation under load be L. Then the strain... for: ; The wind speed measured using an anemometer is V. w The wind force F is calculated using the following formula. w : Where ρ is the air density, A is the windward area, and C is the air density. d This is the drag coefficient.

3. The bridge load intelligent management method based on artificial intelligence according to claim 1, characterized in that: The S2: data preprocessing stage further includes the following steps: Data cleaning is performed by identifying and removing outliers caused by equipment failure, transmission errors, or human interference using the Z-score method. Data standardization is achieved through the min-max normalization method, enabling data from different sources and types to be compared and analyzed on the same scale. When the absolute value of a data point's Z-score exceeds a set threshold, the data point is considered an outlier. The formula for calculating the Z-score is: , where x is the value of the data point, μ is the mean of the sample set, and σ is the standard deviation of the sample set; if |Z| of a data point > 3, then the data point is an outlier and should be removed. The formula for data standardization using the min-max normalization method is as follows: , where X norm The dataset is the normalized dataset, where X is the original data. min X is the minimum value in the dataset. max This represents the maximum value in the dataset.

4. The bridge load intelligent management method based on artificial intelligence according to claim 1, characterized in that: The S3 model training phase also includes the following steps: The prediction model identifies key factors affecting bridge load and predicts bridge load response or bridge health status. During training, the accuracy of the prediction model is improved through loss functions and optimization algorithms, and cross-validation is used to enhance the generalization ability of the prediction model. The cleaned, formatted, and standardized dataset The data points are input into the prediction model, where each data point represents an observation sample. Through correlation analysis or internal model weight evaluation, the features with the greatest impact on bridge load prediction are selected. For dataset X... norm Assuming p features are selected, the input feature matrix is: ,in This represents the p-th feature of the n-th observation sample; When training a prediction model, a neural network is selected to perform a nonlinear mapping from the input features X. input To the predicted value Training, predicted values Generate based on the prediction target of bridge load impact: ,in θ is the activation function of the neural network; θ represents the model parameters, indicating the weights and biases of the neural network. The model parameters are optimized by minimizing the loss function. If the training objective is to minimize the mean squared error, then the loss function is: ,in y is the predicted bridge load value for the i-th sample predicted by the prediction model. i is the actual value of the bridge load for the i-th sample, and N is the number of samples; We optimize using the gradient descent algorithm, updating the model parameters θ to minimize the loss function. Where η is the learning rate, It is the gradient of the loss function with respect to the model parameters θ; The test is performed using a validation set, and the final evaluation result is the average of all validation results. ,in It is the loss function of the i-th validation set; Regularization is used to limit the complexity of the prediction model and prevent overfitting. The L2 regularization loss function is: , where λ is the regularization coefficient and m is the number of model parameters θ.

5. The bridge load intelligent management method based on artificial intelligence according to claim 1, characterized in that: The S4: Management Optimization phase also includes the following steps: By combining and analyzing real-time monitoring data and model prediction results, traffic flow control strategies are adjusted to optimize bridge load distribution, reduce uneven bridge loads, and avoid bridge overloading. Based on the load response predicted by the model, the response state of the bridge structure under different load conditions is analyzed. Let the response of the bridge be y. bridge The prediction model then uses the vehicle load data x as input. load and environmental condition data x env Predicting the bridge's response: ,in This represents the mapping of the load output by the prediction model to the bridge structure; Using the bridge response results output by the predictive model, optimize traffic flow and vehicle speed, and set the objective function. To minimize the maximum stress or deformation of the bridge and avoid local overload, an objective function is set. A comprehensive evaluation standard for the load response of bridges: ,in It is the predicted response value of the i-th bridge monitoring point. This is the current traffic flow configuration; Based on the objective function Generate a traffic flow control strategy to minimize the load on the bridge by optimizing traffic flow and vehicle speed. Define the optimization problem as follows: .

6. The bridge load intelligent management method based on artificial intelligence according to claim 1, characterized in that: The S5: Continuous Monitoring phase also includes the following steps: The bridge's health status data is collected in real time using sensors, and the bridge status data is set as follows: Where m1 represents the number of sensors monitored in real time, y m1 This represents the real-time observation value of the m1-th sensor; Will The input is fed into the prediction model for real-time prediction. Let the output of the prediction model be... The predicted load response of the bridge under its current condition is calculated using the following formula: ,in For the prediction function of the prediction model; When the bridge's monitoring points detect overload, the system automatically adjusts traffic flow or speed limits to reduce bridge load. Let the bridge's maximum load response be... The calculation formula is: The traffic flow control strategy is dynamically updated by adjusting the current traffic flow configuration: ,in For the new traffic flow control strategy The adjusted traffic flow configuration is represented by α, where α is the adjustment coefficient. For the predicted maximum bridge load response, This represents the actual monitored maximum bridge load response.

7. The bridge load intelligent management method based on artificial intelligence according to claim 1, characterized in that: The S6 feedback iteration phase further includes the following steps: The adjusted traffic flow control strategy, bridge load response data, bridge health status before and after the adjustment, and traffic flow conditions are recorded and represented as follows: ,in This represents the real-time data of the nth sensor after the t-th adjustment, where n is the number of sensors detected in the adjustment strategy; After recording the results of each adjustment, the bridge load, stress, and location parameters before and after the adjustment were compared to assess whether the adjusted traffic flow control strategy effectively reduced the bridge load and to analyze whether the adjustment measures had a positive impact on bridge health. Error analysis was used to quantitatively evaluate the adjustment effect, comparing predicted values ​​with actual values ​​and calculating the mean absolute error. for: ,in It is a predicted value; These are actual observed values; Mean absolute error, representing the prediction accuracy of each adjustment measure; Based on the feedback information of the adjustment effect, the parameters of the prediction model are improved through incremental learning, and the prediction model is optimized through gradient descent algorithm during the update. Based on the updated prediction model, combined with the mean absolute error Further optimize traffic flow control strategies; After each round of feedback and optimization, new predictive models and traffic flow control strategies will be iterated.

8. An intelligent bridge load management system based on artificial intelligence, characterized in that: The method for implementing the intelligent bridge load management method based on artificial intelligence as described in any one of claims 1-7 includes: The data acquisition module is used to collect vehicle information, bridge information, and environmental information using various monitoring devices to obtain the first bridge data; The data preprocessing module is used to preprocess the first bridge data to obtain the second bridge data; The model training module is used to train an artificial intelligence algorithm model using the second bridge data to obtain a prediction model; by analyzing real-time bridge data through the prediction model, the impact of dynamic bridge loads on the bridge structure is predicted, and potential risks are identified in advance to obtain analysis results. The management optimization module is used to automatically adjust the traffic flow control strategy for bridges based on the analysis results; The continuous monitoring module is used to continuously monitor changes in the bridge's condition and update the prediction model and traffic flow control strategy based on these changes. The feedback iteration module records the results and feedback information of each adjustment to the traffic flow control strategy, and continuously iterates the prediction model based on the feedback information.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, which, when loaded and executed by a processor, implement the intelligent bridge load management method based on artificial intelligence as described in any one of claims 1-7.

Images