Urban traffic signal control method, device, equipment and medium

By integrating multi-source traffic data and utilizing graph attention networks and deep learning models, intersection state representations and future traffic states are generated, solving the problems of adaptability and decision-making dimensions of traffic signal control in complex dynamic traffic flows, and realizing dynamic adjustment and control optimization of green light duration.

CN121617264APending Publication Date: 2026-03-06CHONGQING ZHILU YUNXING TECHNOLOGY CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202511804734.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-12-03
Publication Date
2026-03-06

AI Technical Summary

Technical Problem

Existing traffic signal control methods are not adaptable to complex dynamic traffic flows. The mismatch between phase and duration decision frequencies leads to control conflicts and oscillations, and prediction errors affect control reliability.

Method used

By acquiring multi-source traffic data from various intersections in the city, we use graph attention networks and multi-head attention mechanisms to generate intersection state representations. We combine bi-branch value assessment deep networks to calculate the optimal phase and green light duration, use a spatiotemporal traffic flow predictor to predict future traffic conditions, and integrate them into control commands for dynamic adjustment.

Benefits of technology

It enhances the adaptability and decision-making dimensions of traffic signal control, enables dynamic adjustment of green light duration, and improves the operational efficiency and control effectiveness of the traffic network.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121617264A_ABST
    Figure CN121617264A_ABST
Patent Text Reader

Abstract

The embodiment of the invention discloses an urban traffic signal control method, device and equipment and a medium, and the method comprises the steps: obtaining multi-source traffic data of all intersections in a city, and carrying out the integration and convergence of the multi-source traffic data to form fused traffic data; generating intersection state representation containing road network context according to the fused traffic data, performing information interaction and competition relation modeling among phases through a multi-head attention mechanism to extract enhanced representation of each phase, and calculating action value of each phase based on the intersection state representation and the enhanced representation of the phase, selecting the phase with the highest action value as the optimal phase; acquiring historical traffic state data, predicting lane-level traffic states of a whole road network in a future preset time window, integrating the lane-level traffic states according to intersections to obtain enhanced state features, and inputting the enhanced state features into a near-end strategy optimization algorithm to obtain an optimal green light duration; and integrating the optimal phase and the optimal green light duration into a control instruction.
Need to check novelty before this filing date? Find Prior Art