The invention discloses an unsupervised method for solving a parameterized partial differential equation based on a graph neural network, and relates to the technical field of solving optimal route recommendation of a parking lot. According to the method, traffic flow, parking space information and path information of a parking lot are collected to construct a parking lot traffic map. And establishing a feature extraction model of the graph convolutional neural network, inputting a final traffic map, solving and calculating the optimal path cost from a certain node to a target parking space through the graph neural network, generating a first optimal path cost value Z, and evaluating to obtain parking space recommendation of the optimal path of the parking lot to obtain a first strategy or a second strategy. When a vehicle is collected to drive out of the parking space in the traffic map, a first optimal path cost value Z or a second optimal path cost value Z2 of the jth vehicle entering the garage is regenerated, an optimization coefficient Yhxs is obtained through calculation, a self-supervision loss function model is constructed through the optimization coefficient Yhxs, an unsupervised learning method is adopted for training, and optimal path recommendation is output.

    本发明公开了一种基于图神经网络求解参数化偏微分方程的无监督方法,涉及求解停车场最优路线推荐技术领域。该方法通过采集停车场的车流量、车位信息和路径信息,用于构建停车场交通图。建立图卷积神经网络的特征提取模型,输入最终交通图,通过图神经网络求解计算从某一节点到目标停车位的最优路径成本,生成第一最优路径成本值Z,并评估得出停车场最优路径停车位推荐,获得第一策略或第二策略。当采集到有车从交通图车位中驶出时,重新生成第j个入库车辆的第一最优路径成本值Z或第二最优路径成本值Z2,计算获得优化系数Yhxs,通过优化系数Yhxs构建自监督损失函数模型,采用无监督学习的方法进行训练,输出最优路径推荐。


    Zugriff

    Download


    Exportieren, teilen und zitieren



    Titel :

    Unsupervised method for solving parameterized partial differential equation based on graph neural network


    Weitere Titelangaben:

    一种基于图神经网络求解参数化偏微分方程的无监督方法


    Beteiligte:
    QIN JIALE (Autor:in) / YI SHICHAO (Autor:in) / DONG SHAN (Autor:in)

    Erscheinungsdatum :

    21.03.2025


    Medientyp :

    Patent


    Format :

    Elektronische Ressource


    Sprache :

    Chinesisch


    Klassifikation :

    IPC:    G06N COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS , Rechnersysteme, basierend auf spezifischen Rechenmodellen / G08G Anlagen zur Steuerung, Regelung oder Überwachung des Verkehrs , TRAFFIC CONTROL SYSTEMS / G06F ELECTRIC DIGITAL DATA PROCESSING , Elektrische digitale Datenverarbeitung




    Nonlinear Dimensionality Reduction for Parameterized Partial Differential Equation

    Peng, L. / Mohseni, K. / American Institute of Aeronautics and Astronautics | British Library Conference Proceedings | 2013


    A New Method of Solving the Hamilton-Jacobi Partial Differential Equation

    Yamashita, Y. / Shima, M. | British Library Online Contents | 1998



    Traffic prediction method based on continuous evolution graph neural controlled differential equation

    CHEN LING / WU JIAJIA | Europäisches Patentamt | 2024

    Freier Zugriff