The invention discloses a short-time traffic flow prediction method based on a CLA-BLSA model, and provides a CLA-BLSA model fusing Conv-LSTM, Bi-LSTM and an attention mechanism, the time-space characteristics and complex nonlinear characteristics of traffic flow are fully mined, and the traffic flow prediction precision is improved. The model comprises a spatial-temporal feature extraction module, a traffic flow date feature extraction module and a traffic flow periodic feature extraction module. The CLA module is a Conv-LSTM module with an attention mechanism, and the module is used for extracting the spatial-temporal characteristics of the traffic flow. Two BLSA modules are designed to be used for capturing daily and weekly periodic change characteristics of a traffic flow, and each BLSA module is composed of a Bi-LSTM module based on a self-attention mechanism. And finally, optimizing the model by adopting a Lookahead optimization algorithm in a manner of assigning an internal loop optimizer and updating a fast weight and a slow weight. According to the method, the Conv-LSTM module of the attention mechanism is introduced to capture the time-space characteristics of the traffic flow, and compared with other existing prediction methods, the method has better prediction performance.

    本发明公开了基于CLA‑BLSA模型的短时交通流预测方法,本方法提出了一种融合了Conv‑LSTM、Bi‑LSTM和注意力机制的CLA‑BLSA模型,充分挖掘交通流的时空特性和复杂非线性特征,提升交通流预测的精度。该模型包含时空特征提取模块、交通流量日期性特征提取模块和交通流量周期性特征提取模块三个模块。CLA模块是带有注意力机制的Conv‑LSTM模块,该模块用于提取交通流的时空特征。设计两个BLSA模块用于捕获交通流每日和每周的周期变化特性,BLSA模块由基于自注意力机制的Bi‑LSTM模块构成。最后采用Lookahead优化算法通过指定内部循环优化器,更新快速权重、慢速权重的方式对模型进行优化。本发明通过引入注意力机制的Conv‑LSTM模块捕获交通流的时空特性,与其他现有的预测方法对比,本发明具有更优越的预测性能。


    Zugriff

    Download


    Exportieren, teilen und zitieren



    Titel :

    Short-term traffic flow prediction method based on CLA-BLSA model


    Weitere Titelangaben:

    基于CLA-BLSA模型的短时交通流预测方法


    Beteiligte:
    SONG YONGDUAN (Autor:in) / YAO DONG (Autor:in) / XIANG QING (Autor:in) / FENG BINGZHU (Autor:in) / LONG HONG (Autor:in) / HU FANG (Autor:in) / HUANG QINGYUE (Autor:in) / XU SHUXIN (Autor:in) / WANG YUJUAN (Autor:in)

    Erscheinungsdatum :

    2023-07-18


    Medientyp :

    Patent


    Format :

    Elektronische Ressource


    Sprache :

    Chinesisch


    Klassifikation :

    IPC:    G08G Anlagen zur Steuerung, Regelung oder Überwachung des Verkehrs , TRAFFIC CONTROL SYSTEMS



    Event-Based Short-Term Traffic Flow Prediction Model

    Head, K. L. / National Research Council / Transportation Research Board | British Library Conference Proceedings | 1995


    Event-Based Short-Term Traffic Flow Prediction Model

    Head, K.Larry | Online Contents | 1995


    Long-term and short-term traffic flow prediction model and method

    QU LICHENG / LYU JIAO / WANG HAIFEI et al. | Europäisches Patentamt | 2020

    Freier Zugriff

    Short-term traffic flow prediction method based on three-stage model

    KONG DEHUI / QIU PENGFEI / WANG SHAOFAN et al. | Europäisches Patentamt | 2021

    Freier Zugriff

    Short-term traffic flow prediction method based on Conv1D-LSTM model

    ZHANG ZHIPENG / LIU YUHANG / DAI LEI et al. | Europäisches Patentamt | 2023

    Freier Zugriff