The invention relates to a vehicle speed prediction method and system applied to an urban scene, and the method employs a personalized federal learning method to achieve the personalized aggregation on a server, captures the specific information of a client, and carries out the personalized vehicle speed prediction through a sequence-to-sequence long-short-term memory network model based on multi-head attention. The personalized federal learning method comprises the following steps: a server sends a global model to a client to initialize a local model; the client executes iterative training and uploads the model to the server; the server aggregates the local model to update the global model, and determines whether to calculate a personalized aggregation weight or not according to the number of iterations; and if calculation is needed, obtaining a personalized aggregation weight based on a weighted mean square error, and further generating an aggregation model customized by the client and sending the aggregation model. The network model adopts an encoder-decoder architecture, comprises two layers of long and short-term memory networks and multiple attention layers, and is finally connected with a full connection layer. Personalized vehicle speed prediction can be realized while data privacy is protected.

    本发明涉及一种应用于城市场景的车辆速度预测方法及系统,该方法利用个性化联邦学习方法在服务器上实现个性化聚合,捕获客户端特定信息,并通过基于多头注意力的序列到序列长短期记忆网络模型进行个性化车速预测。个性化联邦学习方法包括:服务器发送全局模型至客户端初始化本地模型;客户端执行迭代训练并上传模型至服务器;服务器聚合本地模型更新全局模型,并根据迭代轮数决定是否计算个性化聚合权重;若需计算,则基于加权均方误差得到个性化聚合权重,进而生成客户端定制的聚合模型并发送。网络模型采用编码器‑解码器架构,包含两层长短期记忆网络和多头注意力层,最后接全连接层。本发明能在保护数据隐私的同时实现个性化车速预测。


    Zugriff

    Download


    Exportieren, teilen und zitieren



    Titel :

    Vehicle speed prediction method and system used in urban scene


    Weitere Titelangaben:

    一种用于城市场景中的车辆速度预测方法及系统


    Beteiligte:
    ZHOU PENGZHAN (Autor:in) / HE YUEPENG (Autor:in) / ZHAI YIJUN (Autor:in) / QIN ZHIDA (Autor:in) / LI MINGYAN (Autor:in) / CHEN CHAO (Autor:in) / GUO SONGTAO (Autor:in)

    Erscheinungsdatum :

    06.12.2024


    Medientyp :

    Patent


    Format :

    Elektronische Ressource


    Sprache :

    Chinesisch


    Klassifikation :

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



    Urban road vehicle speed prediction method

    FAN ZIDE / HUANG FEILONG / SUN YI et al. | Europäisches Patentamt | 2023

    Freier Zugriff

    Vehicle speed prediction method and system under dynamic vehicle following scene considering traffic environment

    ZHANG ZHONGQUAN / XUE QIFAN / ZHAO XUAN et al. | Europäisches Patentamt | 2025

    Freier Zugriff

    Expressway short-time vehicle speed prediction method considering dynamic traffic scene

    SUO HUANZHI / WANG CHUN / ZHANG WEIHUA et al. | Europäisches Patentamt | 2024

    Freier Zugriff

    Vehicle control method and system used in highway scene

    ZUO GUOMIAO / LEI YONGFU / CHUNG WON-JIN et al. | Europäisches Patentamt | 2023

    Freier Zugriff

    Dangerous scene prediction device, dangerous scene prediction method, and dangerous scene prediction program

    FUTATSUGI YASUNORI / MISHIMA YOSHIHIRO / FUKUZATO ATSUSHI et al. | Europäisches Patentamt | 2025

    Freier Zugriff