The present invention relates to a method and a device for predicting road network travel speed by using a deep neural network. The travel speed in a given road segment is affected by the current and past travel speed of adjacent road sections, and the influence is further extended to the rest of a traffic network. Therefore, a successful prediction model must consider not only the influence of the adjacent road sections but also the influence of distant road sections. Based on the principle, a deep neural network structure for receiving not only time dependence but also spatial correlation across the city is proposed for the topology of an actual traffic network, and an extended model of the proposed prediction model is proposed in terms of traffic state transition and propagation. The present invention was carried out by using a large data set collected for more than 10 months, and successfully predicted the travel speed of 170 road sections in Gangnam, Seoul. The method comprises the following steps of: forming a connection between layers or neurons, and inputting a travel speed data set to an input layer of the deep neural network; obtaining an estimate of a future travel speed for a road network, and measuring an error by comparing the estimated travel speed with an actual observed value by using a loss function; and modifying a learning parameter of the deep neural network according to the error, and improving the prediction accuracy of the travel speed of the road network.

    본 발명은 심층 신경망을 이용한 도로망 통행속도 예측 방법 및 장치에 관한 것이다. 주어진 도로 구간의 통행속도는 인근 도로 구간들의 현재 및 과거 통행속도의 영향을 받으며, 그 영향은 교통망의 나머지 부분으로 더 확장된다. 따라서 성공적인 예측 모델은 이웃의 영향뿐만 아니라 멀리 있는 도로 구간의 영향도 고려해야 한다. 이 원리를 바탕으로, 실제 교통망의 토폴로지를 시간 의존성뿐만 아니라 도시 전역의 공간적 상관관계를 수용하기 위한 심층 신경망 구조를 제안하며, 제안된 예측 모델을 통행상태 전이 및 전파 측면에서 확장한 모델을 제시한다. 본 발명은 10개월 이상 동안 수집된 대규모 데이터 세트를 사용하여 수행되었으며, 서울 강남의 170개 도로 구간의 통행속도를 성공적으로 예측했다.


    Access

    Download


    Export, share and cite



    Title :

    Traffic speed prediction using a deep neural network to accommodate citywide spatio-temporal correlations


    Additional title:

    심층 신경망을 이용한 도로망 통행속도 예측 방법 및 장치


    Contributors:
    LEE YONG JIN (author) / SOHN KEE MIN (author)

    Publication date :

    2020-07-13


    Type of media :

    Patent


    Type of material :

    Electronic Resource


    Language :

    Korean


    Classification :

    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