The invention discloses a space-time prediction method fusing adaptive selection convolution receptive fields. The core of the space-time prediction method is to introduce an ARFSLSTM module. The ARFSLSTM module fuses an adaptive selection receptive field ARFS module and a convolutional long and short term memory model ConvLSTM, extracts context spatio-temporal information through combined action, and performs accurate modeling on correlation of historical states. And the ARFS module is composed of three parts, namely, Split, Fuse and Select. Through a non-linear method, the ARFS module can aggregate information from a plurality of cores, so that each neuron can adaptively adjust the size of a receptive field based on a plurality of input information scales, thereby realizing fine capture of spatio-temporal data. By connecting a recurrent neural network (RNN), a complete spatio-temporal data modeling network is constructed. The network can deeply capture time-space relations and features in a training stage. In conclusion, by introducing the ARFSLSTM module, accurate capture and prediction of the spatio-temporal data are realized, the uniqueness, innovativeness and superiority of the spatio-temporal data are shown, and a new way is opened up for development of the spatio-temporal prediction field.

    一种融合自适应选择卷积感受野的时空预测方法,其核心在于引入ARFSLSTM模块。ARFSLSTM模块融合了自适应选择感受野ARFS模块与卷积长短期记忆模型ConvLSTM,共同作用提取上下文时空信息,并对历史状态的相关性进行精准建模。ARFS模块由Split、Fuse和Select三部分构成。通过非线性方法,ARFS模块能够聚合来自多个核的信息,使每个神经元能够基于多个输入信息尺度自适应地调整其感受野大小,从而实现对时空数据的精细捕捉。通过连接循环神经网络RNN,本发明构建了完整的时空数据建模网络。这一网络在训练阶段能够深入捕捉时空关系和特征。综上所述,本发明通过引入ARFSLSTM模块,实现了对时空数据的精准捕捉和预测,展现了其独特性、创新性和优越性,为时空预测领域的发展开辟了新的道路。


    Access

    Download


    Export, share and cite



    Title :

    Space-time prediction method fusing adaptive selection convolution receptive field


    Additional title:

    一种融合自适应选择卷积感受野的时空预测方法


    Contributors:
    SHI BENYUN (author) / GE CONGHUI (author)

    Publication date :

    2024-07-30


    Type of media :

    Patent


    Type of material :

    Electronic Resource


    Language :

    Chinese


    Classification :

    IPC:    G06N COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS , Rechnersysteme, basierend auf spezifischen Rechenmodellen / G01W METEOROLOGY , Meteorologie / G06F ELECTRIC DIGITAL DATA PROCESSING , Elektrische digitale Datenverarbeitung / G06V / G08G Anlagen zur Steuerung, Regelung oder Überwachung des Verkehrs , TRAFFIC CONTROL SYSTEMS



    Traffic flow prediction method fusing one-dimensional convolution and attribute enhancement unit

    LI YONGQIANG / LI JINYE / FENG YUANJING et al. | European Patent Office | 2023

    Free access

    Traffic flow prediction method and system based on image convolution circulation network fusing Transform

    ZHANG CHEN / WU YUE / ZHANG XIN et al. | European Patent Office | 2024

    Free access

    Track generation method fusing space-time correlation information

    TU LAI / DING ZEYU / KANG YING | European Patent Office | 2024

    Free access

    Traffic flow prediction method based on adaptive lightweight time convolution network

    ZHANG SHUAI / YIN XIANG / ZHANG WENYU et al. | European Patent Office | 2024

    Free access

    Traffic prediction method based on dual dynamic space-time diagram convolution

    SUN YANFENG / JIANG XIANGHENG / HU YONGLI et al. | European Patent Office | 2022

    Free access