基於T-GCN深度學習之台灣市區交通流量預測與視覺化

Traffic Flow Prediction and Visualization in Urban Areas of Taiwan Based on T-GCN Deep Learning Method

蕭立翎(Li-Ling Hsiao)

指導教授 ： 高君豪

淡江大學/商管學院/統計學系數據科學碩士班/碩士(2023年)

https://doi.org/10.6846/tku202300431

摘要

預測道路交通速度一直都是智能交通系統(ITS)熱門的研究話題。不僅可以為駕駛者提前提供交通資訊，節省行車時間；也可以為政府對交通的管理提出建議，例如交通信號燈的控制，擴建道路等。
現有的交通預測方法大部分只考慮空間依賴性或時間依賴性，本研究基於T-GCN的深度學習方法對臺北市信義區進行道路交通速度的預測，利用圖卷積網絡(GCN)和門控循環單元(GRU) 來獲取道路的空間特徵和數據的時間特徵，並且將預測結果以不同的路線顔色呈現在OpenStreetMap的地圖上。

關鍵字

交通預測 ； 深度學習 ； 道路路線視覺化

並列摘要

Road traffic speed forecasting has become a popular research topic in intelligent transportation systems (ITS). It not only provides drivers with advance traffic information, saving them time, but also advises the government on traffic management, such as traffic light control or road construction.
Most existing traffic prediction methods only consider either spatial dependence or temporal dependence. This research is based on the deep learning method of T-GCN to predict road traffic speed in Xinyi District, Taipei City. It utilizes the graph convolutional network (GCN) and the gated recurrent unit (GRU) to capture both the spatial features of the road and the temporal features of the data. Finally, the prediction results are presented on the OpenStreetMap map with different route colors.

並列關鍵字

Traffic Forecast ； Deep Learning ； Road Route Visualization

參考文獻

參考文獻

1. Ahmed, M. S., Cook, A. R. (1979). Analysis Of Freeway Traffic Time-Series Data By Using Box-Jenkins Techniques (No. 722).

https://trid.trb.org/view/148123

2. Lee, S., Fambro, D. B. (1999). Application Of Subset Autoregressive Integrated Moving Average Model For Short-Term Freeway Traffic Volume Forecasting. Transportation research record, 1678(1), 179-188.

https://journals.sagepub.com/doi/full/10.3141/1678-22?casa_token=D9xMc2O3kk4AAAAA:gPOzs-sldPiYybr2T0GQWNztUWOEVysMTYnHMA9GOA4mKFTU2nS0WviBPtDmyqE6uvm9vZs5x2cw8g