Track traffic mode classification method and device, electronic equipment and storage medium

By introducing a comparative learning framework and dynamic queue mechanism in traffic pattern classification, the high-dimensional features of trajectory data are extracted and its time-space dependencies are modeled, and the existing methods are solved in the shortcomings of processing high-dimensional complex trajectory data, achieving higher classification accuracy and robustness.

CN119740101AActive Publication Date: 2025-04-01QILU UNIVERSITY OF TECHNOLOGY (SHANDONG ACADEMY OF SCIENCES) +2
View PDF 5 Cites 2 Cited by

Patent Information

Application Number
CN202510245695.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-03-04
Publication Date
2025-04-01
Estimated Expiration
2045-03-04

AI Technical Summary

Technical Problem

The existing traffic pattern classification methods have insufficient utilization of global information and timing dependence when processing high-dimensional complex trajectory data, and have limited performance in small sample scenarios and complex classification tasks.

Method used

A new contrast learning framework is proposed to construct a multi-layer convolutional network and a Transformer encoder to extract high-dimensional feature representations of trajectory data through deep neural networks, and to model the time series dependencies of trajectories in combination with Gated Recurrent Unit (GRU). This framework innovatively introduces a dynamic queue mechanism, dynamically expanding the number and diversity of negative samples, improving the discriminant ability of feature representation and the classification performance of the model.

Benefits of technology

It significantly enhances the classification accuracy and robustness of the model, can effectively capture the complex spatio-temporal relationships of trajectory data, meets the needs of trajectory data analysis in complex scenarios, and provides efficient and reliable technical solutions for intelligent transportation, urban planning and dynamic traffic management.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119740101A_ABST
    Figure CN119740101A_ABST
Patent Text Reader

Abstract

The invention belongs to the technical field of intelligent traffic, and more particularly relates to a track traffic mode classification method and device, electronic equipment and a storage medium. The method comprises the following steps: performing earlier-stage processing operation on original trajectory data to construct a trajectory data target set; inputting the trajectory data target set as training data into a multi-stage trajectory feature encoder, and training the multi-stage trajectory feature encoder by adopting a comparative learning method in combination with an autoregression model; and extracting high-dimensional feature representation of trajectory data by using the trained multi-stage trajectory feature encoder, and using the feature representation as input for training an MLP classifier, so that the MLP classifier learns a mapping relation between trajectory features and traffic modes, and finally, the MLP classifier is used for trajectory traffic mode classification. According to the method, efficient and accurate traffic mode classification is realized, application requirements in complex scenes are met, and reliable technical support is provided for the fields of intelligent traffic, urban planning, dynamic traffic management and the like.
Need to check novelty before this filing date? Find Prior Art

Citation Information

Patent Citations

  • Strip mine truck scheduling method based on road network-track combined comparative learning

    CN116629531A

  • Track similarity measurement method based on comparative predictive coding, computer readable storage medium and computer program product

    CN119046698A

  • Intelligent automobile multi-mode non-autoregressive trajectory prediction method and device and medium

    CN119293749A

  • Deep learning method for multiple object tracking from video

    US20240144489A1

  • Massive high-dimensional AIS trajectory data clustering method

    WO2023029461A1

Cited By

  • Traffic flow prediction method based on self-supervised spatio-temporal representation and scene adaptation

    CN121708754A

  • Traffic flow prediction method based on self-supervised spatio-temporal representation and scene adaptation

    CN121708754B