Emergency lane control method based on multi-model fusion short-term traffic prediction

Through the multi-model prediction method that integrates GNN, LSTM and ARIMA models, the problem of difficult to capture spatial and time dependence in traffic flow prediction is solved, and high-precision short-term traffic prediction is achieved, providing a scientific basis for emergency lane control and alleviating traffic congestion.

CN120472669APending Publication Date: 2025-08-12ANHUI TRANSPORT CONSULTING & DESIGN INST
View PDF 6 Cites 1 Cited by

Patent Information

Application Number
CN202510852074.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-06-24
Publication Date
2025-08-12

AI Technical Summary

Technical Problem

The existing traffic flow prediction methods are difficult to capture the spatial dependence and time dependence in the traffic network at the same time, resulting in the inability to effectively alleviate traffic congestion.

Method used

A multi-model prediction method combining GNN, LSTM and ARIMA models is adopted, combining spatial feature extraction, time series modeling and trend capture capabilities, a multi-modal fusion model is built for short-term traffic flow prediction, and the model weight is adjusted through Bayesian optimization to improve prediction accuracy and calculation efficiency.

Benefits of technology

It significantly improves the accuracy and computing efficiency of traffic flow prediction, provides scientific emergency lane control decisions, efficient use of road resources, and alleviates traffic congestion.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120472669A_ABST
    Figure CN120472669A_ABST
Patent Text Reader

Abstract

The invention discloses an emergency lane control method based on multi-model fusion short-term traffic prediction, and belongs to the technical field of intelligent traffic systems. According to the method, the advantages of the GNN model, the LSTM model and the ARIMA model are fused, so that the spatial dependency, the time dependency and the trend characteristics of data in the traffic network can be captured at the same time, and the traffic flow prediction precision is remarkably improved; by optimizing the weight of the fusion model, on the premise of guaranteeing the prediction precision, the calculation efficiency of the model is improved, the requirement for real-time decision making in traffic management is met, and based on a high-precision short-time traffic prediction result, a scientific basis can be provided for dynamic opening and closing decision making of an emergency lane, and the method is suitable for popularization and application. Therefore, road resources are utilized more efficiently, and traffic jam is relieved. The fusion model has strong generalization ability, can adapt to the influence of complex and changeable traffic flow characteristics and emergencies in a traffic network, and provides more flexible and reliable decision support for a traffic management department.
Need to check novelty before this filing date? Find Prior Art

Citation Information

Patent Citations

  • CNN-LSTM neural network model and ARIMA model-based time sequence prediction method and system

    CN112819136A

  • Method for determining highway lane management and control strategy based on congested road section

    CN113870562A

  • Expressway bottleneck section emergency lane management and control system

    CN117612378A

  • Dynamic emergency lane opening decision-making method, system, equipment and medium

    CN117789469A

  • Distribution network planning index prediction method based on artificial intelligence

    CN119204565A

Cited By

  • Traffic jam long-period prediction method and system fusing spatial-temporal characteristics

    CN121354354A