Digital management system for highway infrastructure

Abstract

The digital management system used for the highway infrastructure comprises a data acquisition unit used for data acquisition; the fusion unit is used for fusing the data acquired by the data acquisition unit to form a standardized data set; the digital base map construction unit is used for constructing a digital base map according to the data fused by the fusion unit; the grading unit is used for determining the weight of each factor and grading; the infrastructure coding and standard specification unit is used for establishing a system by adopting a hexadecimal coding system and evaluating the maturity of the infrastructure based on a fuzzy comprehensive evaluation method; the running condition dynamic monitoring and early warning unit is used for monitoring the traffic running condition and the infrastructure state of the highway in real time by means of cameras arranged in the highway and the tunnel; the digital linkage unit is used for establishing a correlation model between facilities and equipment; the objective of the invention is to solve the problems of lack of digital base map precision standards, insufficient visual means evaluation guidance and lack of facility and equipment digital linkage in highway infrastructure digital management.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent transportation engineering and relates to a digital management system for highway infrastructure. Background Art

[0002] With the vigorous development of intelligent transportation, promoting the deep integration of new technologies such as big data, the Internet, and artificial intelligence with the transportation industry has become a key measure in the strategy of building a transportation power. As an important part of the transportation network, the digital management of highway infrastructure is crucial for improving traffic operation efficiency and ensuring traffic safety. In this context, numerous studies have been dedicated to using advanced technologies to achieve efficient control and optimized operation of highway infrastructure.

[0003] Traditional highway infrastructure management mostly adopts the mode of "facility entity + paper completion data + manual inspection", which has many drawbacks. On the one hand, the digitalization level of infrastructure is generally low, making it difficult to achieve accurate positioning and real-time status monitoring of various facilities. For example, the information acquisition and management of facilities such as traffic signs, guardrails, and milestones are relatively lagging. On the other hand, there is a lack of an effective digital linkage mechanism among facilities, which cannot provide strong support for intelligent highway business scenarios. For example, it is difficult to quickly respond and cooperate in traffic flow control, emergency management, etc., seriously affecting the intelligent management level and service quality of highways.

[0004] In response to the above problems, the patent titled "Digital System for Traffic Safety Facilities Based on Intelligent Internet of Vehicles", patent number: CN112686407A, which covers multiple functional modules such as a multi-source data fusion basic information module and an intelligent Internet of Vehicles digital facility management module. Through the digital Internet of Vehicles module for traffic safety facilities in the field based on GPS / Beidou / LBS positioning technology and mobile wireless communication transmission methods to collect data, and integrating manually entered and relevant traffic system interaction data, it realizes the comprehensive management and monitoring of traffic safety facilities. This patent system can achieve multi-source data fusion, conduct full-life cycle management of traffic safety facilities, and effectively improve the intelligent level of facility management. For example, in the aspect of intelligent Internet of Vehicles digital facility management, it can achieve functions such as status monitoring, information transmission, and system management for traffic signs, guardrails, etc.; in terms of maintenance assistance support, it can automatically generate maintenance plans to improve maintenance efficiency; in terms of traffic safety evaluation assistance support, it can provide evaluation criteria and track the effects to enhance the scientific nature of evaluation. However, this system mainly focuses on the management of traffic safety facilities and involves relatively less in other aspects of highway infrastructure, such as road surface condition monitoring and bridge structure health monitoring. The breadth and depth of the overall highway infrastructure management still need to be expanded, and there may be deficiencies in adaptability in the complex environments of highways in different regions.

[0005] Based on the existing concept of multi-source data fusion and intelligent management, the present invention aims to further expand the scope of digital management of highway infrastructure. By comprehensively applying high-precision measurement technologies such as vehicle-mounted lidar, GNSS positioning technology, and AI image recognition technology, a digital management system covering multiple dimensions including the research on the classification standard of highway high-definition digital base maps, the evaluation of application maturity, the digital-physical integration application of "construction, management, maintenance, operation, and service" business, and the construction of standards and specifications is comprehensively constructed to achieve panoramic, real-time, and accurate digital management of highway infrastructure, effectively overcome the limitations of existing technologies, and meet the intelligent development needs of highways. Summary of the Invention

[0006] The present invention provides a digital management system for highway infrastructure, which solves the problems that the lack of a unified accuracy standard for digital base maps leads to inaccurate and unreliable data, the lack of clear guidance for the classification and application maturity evaluation of visualization means such as high-definition digital base maps results in unreasonable resource allocation and poor application effects, and the lack of effective digital linkage of highway facilities and equipment, making it difficult to provide strong technical support and efficient collaboration for intelligent highway business scenarios.

[0007] To solve the above problems, the technical solution adopted by the invention is:

[0008] A digital management system for highway infrastructure, comprising:

[0009] A data acquisition unit, the data acquisition unit includes a vehicle-mounted lidar, a GNSS positioning system, a camera, and an Internet of Things sensor;

[0010] The vehicle-mounted lidar is used to collect data on the contour, slope, curvature, position, and shape of obstacles on the road;

[0011] The GNSS positioning system is used to detect data on the position, speed, and driving trajectory of the vehicle;

[0012] The camera is used to detect data on vehicle types, quantities, traffic flow, and the appearance status of facilities;

[0013] The Internet of Things sensor is used to detect data on the vibration frequency of bridges, the stress and strain of tunnels, temperature, humidity, and wind speed;

[0014] A fusion unit, which integrates the data collected by the data acquisition unit through a data fusion algorithm to form a standardized data set. Based on the principle of Bayesian estimation, different data sources are defined as D1, D2, D3... D3, the prior probability is p(D i ), and the likelihood function is p(X|D i ), then the probability distribution of the fused data X is:

[0015]

[0016] The digital base map construction unit constructs a digital base map based on the data fused by the fusion unit, generates 3D models of roads and facilities using 3D modeling algorithms, and combines GNSS positioning technology to achieve geospatial positioning and visual display of the models;

[0017] The grading unit determines the weights of various factors using the analytic hierarchy process, constructs a judgment matrix A, and obtains the weight coefficients by solving the eigenvector W. The calculation formula is AW = μ max W, where μ max is the maximum eigenvalue, and then calculates the digital base map level according to the formula ω i is the weight of the i-th factor, and x i is the quantization value of the corresponding factor;

[0018] The infrastructure coding and standard specification unit adopts a hexadecimal coding system, and the structure is E = F + G + H, where F represents the area code, G represents the facility type code, and H represents the serial number; establishes a system, and evaluates the maturity of the infrastructure based on the fuzzy comprehensive evaluation method, determines the evaluation factor set U = {u1, u2, u3…, u n} and the evaluation grade set V = {v1, v2, v3…, v m}, and passes the fuzzy relation matrix

[0019] R = (γ ij ) m×n

[0020] γ ij represents the membership degree of the factor u i to the evaluation grade v j ; calculates the comprehensive evaluation vector B = A·R = (b1, b2,…b n ) using the weight vector A = (a1, a2,…a m );

[0021] The operation status dynamic monitoring and early warning unit, through the real-time image data of the highway traffic scene collected by the data collection unit, including real-time data of different time periods, different weather conditions, and different traffic conditions, annotates the targets in the images, including vehicle types, traffic signs, road surface damage, street lights, guardrails, tunnel ventilation openings, and the annotation content is the category and bounding box coordinates of the targets;

[0022] Inputs the annotated data set into the trained YOLOv5 model with an attention mechanism, and removes redundant detection boxes through the non-maximum suppression algorithm to obtain the target detection results;

[0023] According to the output of the object detection results, the number of vehicles, the distribution of vehicle types, the number, the recognition of traffic signs, and the traffic operation status information on the highway and in the tunnel are statistically counted in real time. At the same time, the status of infrastructure such as the road surface, guardrails, and lighting is monitored to determine whether there are abnormal conditions such as damage and faults. Based on the vehicle position and speed data obtained from real-time object detection, they are input as the observation values of the Kalman filter algorithm, and the prediction equation and measurement update equation are used for iterative calculation to predict the change trend of traffic flow and speed in the future. According to the real-time monitored traffic operation status and infrastructure status, combined with the prediction results of the Kalman filter algorithm, when the predicted traffic flow exceeds 80% of the road design capacity and the speed continues to decline, it is determined that congestion is imminent, and a traffic congestion warning is issued; when the detected damaged road surface area exceeds a certain threshold or the guardrail is severely deformed, an infrastructure failure warning is issued; when abnormal behaviors such as illegal lane changes and sudden braking of vehicles frequently occur, a traffic accident risk warning is issued, and the warning signal is output;

[0024] An alarm unit, which is used to generate and publish a warning message containing the warning type, location, time, and relevant image evidence.

[0025] The principle of this solution lies in:

[0026] Comprehensive data of highway infrastructure and operation status are obtained through multi-type data acquisition units. The fusion unit integrates multi-source heterogeneous data to form a standardized data set. The digital base map construction unit generates a visual digital base map and classifies it. The infrastructure coding and standard specification unit realizes coding and maturity evaluation. The operation status dynamic monitoring and warning unit uses image data to achieve object detection, traffic and infrastructure status monitoring, traffic flow speed prediction, and multi-type warning judgment, and timely publishes warning messages through the alarm unit, so as to realize the digital management and dynamic monitoring and warning of highway infrastructure and ensure the safe and efficient operation of highways.

[0027] The beneficial effects of this solution:

[0028] Through the collaborative work of multiple sensors, various key data of highway infrastructure can be obtained, making the understanding of highways no longer limited to local or single dimensions. The fusion unit effectively integrates multi-source heterogeneous data through a data fusion algorithm based on the Bayesian estimation principle to form a standardized data set, eliminating the differences and conflicts between different data sources, ensuring the consistency and reliability of the data, providing high-quality data input for subsequent processing and analysis, and reducing the risk of decision-making errors caused by data errors or inconsistencies.

[0029] The visual digital base map generated by the digital base map construction unit combining 3D modeling and GNSS positioning technology provides managers with an intuitive and clear overall view of the highway. At the same time, the grading system enables more targeted allocation of resources, focusing on key sections and facilities, improving the efficiency and accuracy of management.

[0030] The introduction of the infrastructure coding and standard specification unit realizes the standardized coding and maturity assessment of infrastructure through the hexadecimal coding system and the fuzzy comprehensive evaluation method. It is beneficial to improve the standardization and normalization level of management, ensure the accurate identification and scientific evaluation of infrastructure, and provide a strong basis for decisions such as maintenance, update, and expansion.

[0031] Through image annotation, object detection technology, and the Kalman filter algorithm, it is possible to monitor the traffic operation status and infrastructure status in real time and accurately, predict the changing trends of traffic flow and speed, and issue various types of early warnings in a timely manner. This enables the management department to take measures in advance to prevent traffic congestion, infrastructure failures, and traffic accidents, ensuring the safe and smooth flow of the highway and reducing potential economic losses and social impacts.

[0032] Furthermore, the vehicle-mounted lidar in the data acquisition unit adopts a multi-line scanning mode.

[0033] Furthermore, the Bayesian estimation principle in the fusion unit introduces an adaptive adjustment factor to dynamically adjust the prior probability and likelihood function according to the real-time nature and reliability of the data.

[0034] Furthermore, the camera in the operation status dynamic monitoring and early warning unit adopts an AI camera.

[0035] Furthermore, when constructing the 3D model, the digital base map construction and grading unit adopts the Poisson reconstruction algorithm to process large-scale point cloud data and generate a high-quality road and facility surface model; the change detection algorithm is used to identify new or changed facility information.

[0036] Furthermore, in the operation status dynamic monitoring and early warning unit, transfer learning technology is adopted, and a pre-trained model is used in a large-scale traffic image dataset to accelerate the convergence speed of model training.

[0037] Furthermore, in the digital base map construction and grading unit, a generative adversarial network architecture is adopted, and adversarial training is carried out through a generator and a discriminator.

[0038] Furthermore, in the operation status dynamic monitoring and early warning unit, the optimized prediction algorithm is based on a long short-term memory network and is used to process time series data. Description of the Drawings

[0039] Figure 1 Flow chart of the present invention Detailed implementation manners

[0040] Example 1 is basically as shown in the appendix Figure 1 A digital management system for highway infrastructure includes:

[0041] A data acquisition unit, which includes an on-vehicle lidar, a GNSS positioning system, a camera, and an Internet of Things sensor;

[0042] The on-vehicle lidar is used to collect data on the contour, slope, curvature, obstacle position, and shape of the road;

[0043] The GNSS positioning system is used to detect data on the vehicle's position, speed, and driving trajectory;

[0044] The camera is used to detect data on vehicle types, quantities, traffic flow, and the appearance condition of facilities;

[0045] The Internet of Things sensor is used to detect data on the vibration frequency of bridges, the stress and strain of tunnels, temperature, humidity, and wind speed;

[0046] A fusion unit that integrates the data collected by the data acquisition unit through a data fusion algorithm to form a standardized data set. Based on the Bayesian estimation principle, define the data from different data sources as D1, D2, D3... D n , the prior probability is p(D i ), and the likelihood function is p(X|D i ), then the probability distribution of the fused data X is:

[0047]

[0048] A digital base map construction unit that constructs a digital base map according to the data fused by the fusion unit, generates a three-dimensional model of the road and facilities using a three-dimensional modeling algorithm, and combines the GNSS positioning technology to achieve the geographical spatial positioning and visual display of the model;

[0049] A grading unit that determines the weights of various factors using the analytic hierarchy process, constructs a judgment matrix A, and obtains the weight coefficients by solving the eigenvector W. The calculation formula is AW = μ max W, where μ max is the maximum eigenvalue, and then calculates the digital base map level according to the formula , ω i is the weight of the i-th factor, and x i is the quantization value of the corresponding factor;

[0050] Infrastructure Coding and Standard Specification Unit, which adopts a hexadecimal coding system with the structure E = F + G + H, where F represents the area code, G represents the facility type code, and H represents the serial number; establish a system to evaluate the infrastructure maturity based on the fuzzy comprehensive evaluation method, and determine the evaluation factor set U = {u1, u2, u3…, u n} and the evaluation grade set V = {v1, v2, v3…, v m}, and through the fuzzy relation matrix

[0051] R = (γ ij ) m×n

[0052] γ ij represents the membership degree of factor u i to the evaluation grade v j ; the weight vector A = (a1, a2,…a n ) is used to calculate the comprehensive evaluation vector B = A·R = (b1, b2,…b m );

[0053] Operation Status Dynamic Monitoring and Early Warning Unit, which collects real-time image data of highway traffic scenarios through the data collection unit, including real-time data in different time periods, different weather conditions, and different traffic conditions, and labels the targets in the images, including vehicle types, traffic signs, road surface damage, street lights, guardrails, tunnel ventilation openings, and the labeling content is the category and bounding box coordinates of the targets;

[0054] Input the labeled data set into the trained YOLOv5 model with an attention mechanism, and remove redundant detection boxes through the non-maximum suppression algorithm to obtain the target detection results;

[0055] According to the output target detection results, real-time statistics are made on the number of vehicles, vehicle type distribution, quantity, traffic sign recognition situation, and traffic operation status information on the highway and in the tunnel. At the same time, the status of infrastructure such as the road surface, guardrails, and lighting is monitored to determine whether there are abnormal situations such as damage and faults. Based on the data of vehicle positions and speeds obtained from real-time target detection, they are input as the observation values of the Kalman filter algorithm, and the prediction equation and measurement update equation are used for iterative calculation to predict the change trend of traffic flow and speed in the next period of time. According to the real-time monitored traffic operation status and infrastructure status, combined with the prediction results of the Kalman filter algorithm, when the predicted traffic flow exceeds 80% of the road design capacity and the speed continues to decline, it is determined that congestion is imminent and a traffic congestion early warning is issued; when the detected damaged area of the road surface exceeds a certain threshold or the guardrail is severely deformed, an infrastructure fault early warning is issued; when it is found that abnormal behaviors such as illegal lane changes and sudden braking of vehicles frequently occur, a traffic accident risk early warning is issued, and the early warning signal is output.

[0056] An alarm unit for generating and publishing an alarm message containing the warning type, location, time, and relevant image evidence.

[0057] The comprehensive data of highway infrastructure and operating conditions are obtained through a multi-type data acquisition unit. The fusion unit integrates multi-source heterogeneous data to form a standardized data set. The digital base map construction unit generates a visual digital base map and classifies it. The infrastructure coding and standard specification unit realizes coding and maturity evaluation. The operating condition dynamic monitoring and early warning unit uses image data to achieve target detection, traffic and infrastructure status monitoring, traffic flow and speed prediction, and various types of early warning judgments, and publishes early warning messages in a timely manner through the alarm unit, so as to realize the digital management and dynamic monitoring and early warning of highway infrastructure, and ensure the safe and efficient operation of the highway.

[0058] Through the collaborative work of multiple sensors, various key data of highway infrastructure can be obtained, so that the understanding of highways is no longer limited to local or single dimensions. The fusion unit effectively integrates multi-source heterogeneous data through a data fusion algorithm based on the Bayesian estimation principle, forms a standardized data set, eliminates the differences and conflicts between different data sources, ensures the consistency and reliability of data, provides high-quality data input for subsequent processing and analysis, and reduces the risk of decision-making errors caused by data errors or inconsistencies.

[0059] The visual digital base map generated by the digital base map construction unit combining 3D modeling and GNSS positioning technology provides managers with an intuitive and clear overall view of the highway. At the same time, the classification system enables resources to be allocated more targeted, focusing on key sections and facilities, improving the efficiency and accuracy of management.

[0060] The introduction of the infrastructure coding and standard specification unit realizes the standardized coding and maturity evaluation of infrastructure through a hexadecimal coding system and a fuzzy comprehensive evaluation method. It is beneficial to improve the standardization and standardization level of management, ensure the accurate identification and scientific evaluation of infrastructure, and provide a strong basis for decisions such as maintenance, update, and expansion.

[0061] Through image annotation, target detection technology, and the Kalman filter algorithm, the traffic operation conditions and infrastructure status can be monitored in real time and accurately, the change trends of traffic flow and speed can be predicted, and various types of early warnings can be issued in a timely manner. This enables the management department to take measures in advance to prevent traffic congestion, infrastructure failures, and traffic accidents, ensuring the safe and unobstructed operation of the highway and reducing potential economic losses and social impacts.

[0062] The vehicle-mounted lidar in the data acquisition unit adopts a multi-line scanning mode. The multi-line scanning mode can simultaneously obtain data of multiple scanning lines. Compared with single-line scanning, it can cover a larger area in the same time, so as to obtain rich road environment information faster.

[0063] The Bayesian estimation principle in the fusion unit introduces an adaptive adjustment factor, which dynamically adjusts the prior probability and likelihood function according to the real-time performance and reliability of the data, can dynamically adjust the weight according to the real-time performance and reliability of the data, make the fusion result closer to the actual situation, and reduce the error caused by fixed prior probability and likelihood function; at the same time, by flexibly adjusting the prior probability and likelihood function, it can better handle the uncertainty in the data, making the fused result more deterministic and credible.

[0064] The camera in the operation status dynamic monitoring and warning unit adopts an AI camera. The deep learning algorithm built in the AI camera can more accurately identify and distinguish targets, can perform intelligent analysis on the collected images in real time, quickly judge abnormal situations such as traffic congestion and accidents, and issue warnings in time.

[0065] When constructing the 3D model, the digital base map construction and grading unit adopts the Poisson reconstruction algorithm to process large-scale point cloud data and generate a high-quality road and facility surface model; uses the change detection algorithm to identify new or changed facility information. The Poisson reconstruction algorithm can effectively process large-scale point cloud data, overcome the problems of computational efficiency and memory occupation that may be encountered by traditional methods when processing massive data, making it more efficient and fast to construct the 3D model of a large-scale highway, and can generate a smoother, continuous and accurate road and facility surface model, better restoring the real geometric shape and topological structure, and providing a more accurate basis for subsequent analysis and applications.

[0066] In the operation status dynamic monitoring and warning unit, the transfer learning technology is adopted. The pre-trained model is used in a large-scale traffic image dataset to accelerate the convergence speed of model training, avoiding training the model from scratch, reducing the time and computational cost required for training, being able to obtain an available model faster, and at the same time being able to try different model architectures and parameter adjustments faster, accelerating the optimization and innovation of the algorithm to adapt to the ever-changing traffic monitoring and warning requirements.

[0067] In the digital base map construction and grading unit, the generative adversarial network architecture is adopted. Through the adversarial training of the generator and discriminator, through the adversarial training, the model can continuously adapt and optimize, has stronger robustness to noise and abnormal data, and improves the stability of the digital base map.

[0068] In the operating condition dynamic monitoring and early warning unit, the optimized prediction algorithm is based on a long short-term memory network and is used to process time series data. It can well understand the time sequence of the data, so as to more accurately capture the changes in the operating conditions over time. Specific Embodiment 2

[0070] The data acquisition unit starts to work. The vehicle-mounted lidar scans the road in a multi-line scanning mode, quickly and accurately obtaining detailed data such as the road contour, slope, curvature, and the position and shape of obstacles. For example, in a mountainous section, the multi-line scanning mode can simultaneously capture the precise shapes and slope changes of multiple curves.

[0071] The GNSS positioning system real-time tracks the vehicle's position, speed, and driving trajectory. When the vehicle is driving on the highway, the system can accurately record the real-time position and driving path of each vehicle.

[0072] The camera continuously monitors vehicle types, quantities, traffic flow, and the appearance status of facilities. At the toll station of the highway, the camera can clearly distinguish different types of vehicles such as trucks, buses, and cars, and count the number of passing vehicles. At the same time, it can also detect whether there are signs of damage or aging on the appearance of the toll station facilities.

[0073] The Internet of Things sensors then real-time collect data such as the vibration frequency of the bridge, the stress and strain of the tunnel, and the temperature, humidity, and wind speed around the road. For example, on a large bridge, the sensors detect a slight but abnormal change in the vibration frequency of the bridge, and this data is collected and transmitted in a timely manner.

[0074] The collected data is transmitted to the fusion unit. Based on the Bayesian estimation principle, the fusion unit introduces an adaptive adjustment factor, dynamically adjusts the prior probability and likelihood function according to the timeliness and reliability of the data, and integrates these multi-source heterogeneous data into a standardized data set. For example, if the camera data is affected by weather conditions during a certain period, the system will automatically reduce its weight in the fusion to ensure the accuracy and reliability of the fusion result.

[0075] The digital base map construction unit constructs a three-dimensional digital base map based on the fused data using the Poisson reconstruction algorithm. In a complex highway hub area, a high-precision, clear, and intuitive three-dimensional model of the road and facilities is generated, and accurate geospatial positioning and visualization display are achieved by combining GNSS positioning technology. Managers can clearly understand the road layout and facility distribution in this area by viewing the digital base map.

[0076] The grading unit determines the weights of various factors using the analytic hierarchy process, constructs a judgment matrix and calculates the weight coefficients, and then calculates the grade of the digital base map. For sections with high traffic flow and frequent accidents, a higher grade is assigned to give more attention and investment in resource allocation and management.

[0077] The infrastructure coding and standard specification unit codes various types of infrastructure according to the hexadecimal coding system and evaluates their maturity based on the fuzzy comprehensive evaluation method. For example, for a section of highway that has been in service for many years, a comprehensive evaluation is carried out according to factors such as its pavement condition and the degree of facility aging.

[0078] The AI cameras in the operation status dynamic monitoring and early warning unit continuously collect traffic scene images of the highway. When traffic congestion occurs on a section of the highway, the AI cameras can quickly identify it and focus on analyzing the congested area through the attention mechanism in the YOLOv5 model. At the same time, a long short-term memory network accelerated by transfer learning technology is used, with historical traffic flow and speed data as input, and the Kalman filter algorithm is used to predict the change trend of traffic flow and speed in the next period of time.

[0079] If it is predicted that the traffic flow is about to exceed 80% of the road design capacity and the speed continues to decline, the system determines that congestion is imminent and issues a traffic congestion warning; when it is detected that there is a large area of damage to the road surface or the guardrail is severely deformed, an infrastructure failure warning is issued; when it is found that vehicles frequently exhibit abnormal behaviors such as illegal lane changes and sudden braking, a traffic accident risk warning is issued.

[0080] The alarm unit immediately generates a detailed warning message containing the warning type, location, time, and relevant image evidence, and quickly distributes it to relevant management departments and personnel.

[0081] The above are only embodiments of the present invention. Common knowledge such as specific structures and characteristics in the solution is not described in detail here. Those of ordinary skill in the art know all the common general technical knowledge in the technical field to which the invention belongs before the application date or the priority date, can know all the existing technologies in this field, and have the ability to apply conventional experimental means before this date. Those of ordinary skill in the art can, under the inspiration given in this application, combine their own abilities to improve and implement this solution. Some typical well-known structures or well-known methods should not become obstacles for those of ordinary skill in the art to implement this application. It should be noted that for those skilled in the art, without departing from the structure of the present invention, several deformations and improvements can still be made, and these should also be regarded as the protection scope of the present invention, and these will not affect the implementation effect of the present invention and the practicality of the patent. The protection scope required by this application should be subject to the content of its claims, and the specific implementation manners and the like described in the specification can be used to interpret the content of the claims.

Claims

1. A digital management system for highway infrastructure, characterized in that, Including: A data acquisition unit, which includes an on-vehicle lidar, a GNSS positioning system, a camera, and an Internet of Things sensor; The on-vehicle lidar is used to collect data on the contour, slope, curvature, obstacle position, and shape of the road; The GNSS positioning system is used to detect data on the vehicle's position, speed, and driving trajectory; The camera is used to detect data on vehicle types, quantities, traffic flow, and the appearance status of facilities; The Internet of Things sensor is used to detect data on bridge vibration frequency, tunnel stress and strain, temperature, humidity, and wind speed; The fusion unit integrates the data collected by the data acquisition unit through a data fusion algorithm to form a standardized data set. Based on the principle of Bayesian estimation, the data from different data sources are defined as D1, D2, D3…D n , the prior probability is p(D i ), the likelihood function is p(X|D i ), then the probability distribution of the fused data X is: A digital base map construction unit constructs a digital base map based on the data fused by the fusion unit, generates a 3D model of the road and facilities using a 3D modeling algorithm, and combines GNSS positioning technology to achieve model geospatial positioning and visualization display; The grading unit uses the analytic hierarchy process to determine the weights of various factors, constructs a judgment matrix A, and obtains the weight coefficients by solving the eigenvector W. The calculation formula is AW = μ max W, where μ max is the maximum eigenvalue. Then, according to the formula calculate the digital base map level, ω i is the weight of the i-th factor, and x i is the quantization value of the corresponding factor; The infrastructure coding and standard specification unit adopts a hexadecimal coding system, and its structure is E = F + G + H, where F represents the area code, G represents the facility type code, and H represents the serial number; establish a system to evaluate the infrastructure maturity based on the fuzzy comprehensive evaluation method, and determine the evaluation factor set U = {u1, u2, u3…, u n} and the evaluation grade set V = {v1, v2, v3…, v m}, through the fuzzy relation matrix R = (γ ij ) m×n γ ij represents factor u i to the membership degree of evaluation level v j ; weight vector A = (a1, a2, … a m ) to calculate the comprehensive evaluation vector B = A · R = (b1, b2, … b m ); An operating condition dynamic monitoring and warning unit, through the real-time image data of the highway traffic scene collected by the data acquisition unit, including real-time data in different time periods, different weather conditions, and different traffic conditions, annotates the targets in the image, including vehicle types, traffic signs, road surface damage, street lights, guardrails, and tunnel ventilation openings, and the annotation content is the category and bounding box coordinates of the target; The annotated data is input into the trained YOLOv5 model with an attention mechanism, and redundant detection boxes are removed through the non-maximum suppression algorithm to obtain the target detection result; According to the output target detection result, the number of vehicles, vehicle type distribution, quantity, traffic sign recognition situation, and traffic operation condition information on the highway and in the tunnel are statistically analyzed in real time. At the same time, the status of infrastructure such as the road surface, guardrails, and lighting is monitored to determine whether there are abnormal situations such as damage and faults. Based on the data of vehicle position and speed obtained from real-time target detection, they are input as the observation values of the Kalman filter algorithm, and the prediction equation and measurement update equation are used for iterative calculation to predict the change trend of traffic flow and speed. According to the real-time monitored traffic operation condition and infrastructure status, combined with the prediction result of the Kalman filter algorithm, when the predicted traffic flow exceeds 80% of the road design capacity and the speed continues to decline, it is determined that congestion is imminent and a traffic congestion warning is issued; when the detected damaged area of the road surface exceeds a certain threshold or the guardrail shows serious deformation, an infrastructure fault warning is issued; When it is found that abnormal behaviors such as illegal lane changes and sudden braking of vehicles frequently occur, a traffic accident risk warning is issued and the warning signal is output; An alarm unit is used to generate and publish a warning message containing the warning type, occurrence location, time, and relevant image evidence.

2. The digital management system for highway infrastructure according to claim 1, wherein The on-vehicle lidar in the data acquisition unit adopts a multi-line scanning mode.

3. The digital management system for highway infrastructure according to claim 1, characterized in that, The Bayesian estimation principle in the fusion unit introduces an adaptive adjustment factor to dynamically adjust the prior probability and likelihood function according to the real-time nature and reliability of the data.

4. The digital management system for highway infrastructure according to claim 1, characterized in that, The camera in the operating condition dynamic monitoring and warning unit adopts an AI camera.

5. The digital management system for highway infrastructure according to claim 1, characterized in that, When constructing the 3D model, the digital base map construction and grading unit adopts the Poisson reconstruction algorithm to process large-scale point cloud data and generate a high-quality road and facility surface model; uses a change detection algorithm to identify newly added or changed facility information.

6. The digital management system for highway infrastructure according to claim 1, characterized in that, In the operating condition dynamic monitoring and early warning unit, transfer learning technology is adopted, and a pre-trained model is used in a large-scale traffic image dataset to accelerate the convergence speed of model training.

7. The digital management system for highway infrastructure according to claim 1, characterized in that, In the digital base map construction and grading unit, a generative adversarial network architecture is adopted, and adversarial training is carried out through a generator and a discriminator.

8. The digital management system for highway infrastructure according to claim 1, characterized in that, In the operating condition dynamic monitoring and early warning unit, the optimized prediction algorithm is based on a long short-term memory network and is used to process time series data.

Images