Traffic flow optimization model in trusted data space

Abstract

The invention relates to the technical field of intelligent traffic and data trusted computing, and discloses a traffic flow optimization model based on a trusted data space. According to the model, multi-source traffic data including road monitoring equipment, Internet of Vehicles (V2X), unmanned aerial vehicle images and the like are utilized to construct a trusted data space, and data fusion and optimization calculation are realized. And the non-tampering property of the data is ensured through a block chain technology, and a federated learning algorithm is adopted to carry out data analysis to generate an optimal traffic flow scheduling scheme. The optimization strategy comprises intelligent signal lamp control, dynamic lane allocation and intelligent navigation adjustment so as to improve the road traffic efficiency, reduce traffic congestion and improve the intelligent level of traffic management. The method is suitable for intelligent urban traffic management, and can effectively improve the road operation efficiency and reduce the carbon emission.

Description

Technical Field

[0001] The present invention relates to the fields of intelligent traffic optimization and data trusted computing, and in particular to a traffic flow optimization model based on a trusted data space. Background Art

[0002] At present, urban traffic management still faces the following problems: Traditional traffic flow control strategies rely on fixed signal light timings and cannot dynamically adapt to actual traffic conditions, leading to congestion at some intersections. Traffic data sources are scattered, data credibility is low, and effective integration and analysis are difficult; Existing intelligent transportation systems mostly use single-point optimization methods and fail to fully utilize multi-source data for overall optimization.

[0003] In response to the above problems, the present invention proposes a traffic flow optimization model based on trusted data space, which improves the real-time and intelligent level of traffic management through trusted data storage and intelligent optimization calculation. Summary of the Invention

[0004] The purpose of the present invention is to provide a traffic flow optimization model in a credible data space to solve the problems of low data credibility and insufficient optimization calculation efficiency in existing traffic management solutions.

[0005] The technical solutions of the present invention are as follows: Using multi-source data collection, combining vehicle-to-everything (V2X), surveillance cameras, floating vehicle data, and drone images, to achieve wide-area traffic information perception; Maintaining a trusted data space through blockchain technology to ensure data integrity and shared security; Use federated learning algorithms for data privacy protection calculations to achieve cross-departmental data integration and optimized analysis; Combine reinforcement learning methods to optimize traffic signal control and improve intersection efficiency; Adopting a cloud-edge collaborative computing architecture, global optimization is performed in the cloud, and local optimization control is performed at the edge nodes. BRIEF DESCRIPTION OF THE DRAWINGS · Figure 1 Traffic flow optimization system architecture diagram under trusted data space · Figure 2 Schematic diagram of traffic flow optimization calculation process DETAILED DESCRIPTION

[0006] (1) Data acquisition module The module includes vehicle-to-everything (V2X) devices, road sensors, traffic cameras, and drone monitoring systems to collect traffic flow data in real time.

[0007] (2) Trusted Data Space Build a trusted data sharing platform through blockchain technology to ensure the reliability of data sources and prevent data tampering; Adopt smart contracts to achieve automatic verification and distributed storage of traffic data.

[0008] (3) Traffic Optimization Calculation Engine This module adopts a federated learning method, which enables multiple traffic management agencies to jointly train a global optimization model without sharing the original data. By aggregating local training results from various places, the accuracy of traffic flow prediction and optimization decisions can be improved. Each management agency conducts localized training based on locally collected traffic data (including traffic volume, traffic signal status, road conditions, etc.), and periodically uploads updated model parameters to the global model to achieve cross-regional collaborative learning. Through reinforcement learning algorithms, based on the acquisition of real-time traffic status data, the duration of traffic lights is dynamically adjusted, lane allocation plans are optimized, and intelligent navigation path suggestions are provided. Reinforcement learning continuously adjusts traffic control strategies through interaction with the traffic environment to maximize the optimization effect of traffic flow, reduce congestion and delays, and improve road capacity. The specific steps are as follows: Data preprocessing: The system collects real-time traffic data from multiple sources, including vehicle-to-everything (V2X), road sensors, traffic cameras, and drone monitoring systems. After cleaning and standardizing the collected data, valid feature information (such as traffic volume, speed, road conditions, and time period) is extracted, providing high-quality data input for subsequent optimization calculations. Federated learning training process: ① Local model training: Each traffic management agency uses local traffic data to independently train traffic flow prediction and optimization models. Each local model uses historical traffic data in its specific area to perform traffic pattern recognition and flow optimization. ② Model update and aggregation: The update results of each local model are regularly aggregated to the central server through secure aggregation methods (such as weighted averaging, model fusion, etc.) to generate a global optimization model. The global model combines the optimization experience of various regions to improve the accuracy of overall traffic flow scheduling and resource allocation. ③ Global optimization: Through multiple rounds of model updates and iterative training, the global traffic flow management model is continuously optimized. The global model constantly adapts to the traffic needs of different regions, realizing cross-regional and cross-departmental collaborative scheduling and resource sharing. Reinforcement learning decision-making process: ① Environmental modeling: The system designs an environmental model for each intersection and lane, including multi-dimensional information such as current traffic flow, signal status, lane occupancy, and weather conditions. The state of each intersection serves as the "state space" in reinforcement learning. ② Action space: Based on the current traffic state, the system selects and executes possible optimization actions, such as adjusting signal cycles, reallocating lanes, or recommending routes. The execution of each action affects traffic flow and provides feedback to the system. ③ Reward mechanism: Reinforcement learning evaluates the effectiveness of different optimization strategies by designing a reward mechanism. If adjustments to signal durations, lane allocations, and other measures effectively reduce traffic delays and improve traffic efficiency, positive rewards are given; if the solution fails to effectively alleviate traffic congestion, negative rewards are given. Through this reward mechanism, the system can dynamically adjust its strategy based on environmental feedback and gradually learn the optimal traffic flow optimization solution. This module not only handles and optimizes local traffic management but also enables regional collaboration and data sharing at the global level. Combining the privacy-preserving nature of federated learning with the adaptive decision-making capabilities of reinforcement learning, the system can adjust key traffic control strategies like traffic lights, lane assignments, and path planning in real time, achieving intelligent and automated traffic flow optimization. Ultimately, this aims to improve road efficiency, reduce congestion, and lower traffic accident rates and carbon emissions.

[0009] (4) Intelligent control module This module includes intelligent traffic light control, dynamic lane assignment, and intelligent navigation recommendations. The traffic light control submodule uses a reinforcement learning algorithm to adaptively optimize the traffic light cycle. The dynamic lane allocation submodule dynamically adjusts variable lanes according to real-time traffic flow to improve road utilization. The intelligent navigation recommendation submodule combines multi-source data to provide users with real-time route optimization solutions to alleviate traffic pressure in key areas.

[0010] (5) Cloud-edge collaborative computing architecture ·Use cloud-based large-scale computing combined with edge node computing to improve optimization response speed. Enable low-latency communication between the cloud and edge computing devices through the 5G communication network. Summarize

[0011] This paper provides a traffic flow optimization model based on a trusted data space. It leverages multi-source data fusion, trusted blockchain data storage, and federated learning optimization computing to achieve intelligent traffic management optimization. This model can be widely applied in smart cities and traffic management platforms to improve traffic flow efficiency, reduce congestion, and enhance intelligent management.

Claims

1. A traffic flow optimization model in a trusted data space, characterized by: The model includes: Data collection module, which collects real-time traffic data from road monitoring equipment, the V21 connected vehicle system, drone monitoring, and mobile terminals; A trusted data space for storing, sharing, and managing multi-source traffic data, ensuring data credibility through blockchain technology; An optimized computing engine, based on federated learning algorithms and reinforcement learning models, optimizes traffic data and generates optimal traffic flow scheduling strategies; An intelligent control module, which executes optimization results, including dynamic traffic light scheduling, lane adjustment, and intelligent navigation path recommendations.

2. The traffic flow optimization model according to claim 1, characterized in that: The trusted data space uses a blockchain architecture to store data, ensures that the data cannot be tampered with, and supports cross-departmental traffic data sharing.

3. The traffic flow optimization model according to claim 1, characterized in that: The optimization calculation engine uses a federated learning method to perform multi-party traffic flow optimization calculations while protecting data privacy.

4. The traffic flow optimization model according to claim 1, characterized in that: The intelligent control module includes: Intelligent traffic light control submodule, used to dynamically adjust traffic light duration based on predicted traffic flow; Dynamic lane adjustment submodule, used to optimize lane allocation during peak hours and improve traffic efficiency; Intelligent navigation recommendation submodule, used to provide personalized route planning to balance road network traffic.

5. The traffic flow optimization model according to claim 1, characterized in that: The model can be deployed in the cloud and work with edge computing nodes through 5G networks to reduce computing latency and improve system response speed.

Images