| DC 欄位 |
值 |
語言 |
| DC.contributor | 資訊工程學系 | zh_TW |
| DC.creator | 黃鐙毅 | zh_TW |
| DC.creator | Tens-I Huang | en_US |
| dc.date.accessioned | 2019-7-25T07:39:07Z | |
| dc.date.available | 2019-7-25T07:39:07Z | |
| dc.date.issued | 2019 | |
| dc.identifier.uri | http://ir.lib.ncu.edu.tw:444/thesis/view_etd.asp?URN=106522089 | |
| dc.contributor.department | 資訊工程學系 | zh_TW |
| DC.description | 國立中央大學 | zh_TW |
| DC.description | National Central University | en_US |
| dc.description.abstract | 隨著城鄉交通流量的不斷增加，交通信息在智慧型運輸系統中發揮著重要作用。在過去幾年中，許多不同的方法已用於短期交通流量預測。城市交通流量預測有助於為交通壅堵提供早期預警，並作為旅行時間預測的特徵。
本文提出了一種交通流預測系統，用於預測和視覺化桃園市區的小時交通流量。研究使用了從安裝在交叉路口的RFID 車流檢測器收集而來的整年交通數據。為了解決由定期維護或設備故障引起的數據缺失問題，三種不同的缺漏值填補技術被用來清除未校正的數據。兩個基於深度學習的交通流量預測模型，即遞歸神經網路（RNN）和卷積神經網路（CNN），用於預測短期交通流量。最後，設置了一個RShiny在線監平台來演示我們的結果。為了評估所提出的模型的性能，採用了幾個評估指標來驗證我們的模型。實驗結果表明，CNN 模型的準確率在所有模型中表現最佳。 | zh_TW |
| dc.description.abstract | With the steady increasing of both rural and urban traffic flow, traffic information plays an important role in the Intelligent Transport System. Over the last few years, many different methodologies have been used for short-term traffic flow forecasting. The urban traffic flow forecast helps to provide an early warning for traffic congestion and works as a feature for travel time prediction.
In this thesis, we propose a traffic flow prediction system to predict and visualize hourly traffic flow in Taoyuan urban area. A full year traffic data collected from RFID traffic flow detectors installed at the intersections are used in our research. To solve the missing data problem caused by scheduled maintenance or device malfunction, three imputation techniques are carried out to cleanse the uncorrected data. Two deep learning-based traffic flow prediction models, i.e., Recurrent Neural Network and Convolutional Neural Network (CNN), are used to forecast the short-term traffic flow. Finally, a R-Shiny online dashboard is set up to demonstrate our results. To assess the performance of the proposed model, several evaluation metrics are adopted to validate our models. The experiment results show that the accuracy of the CNN model is the highest among all of the models. | en_US |
| DC.subject | 智慧型運輸系統 | zh_TW |
| DC.subject | 車流量 | zh_TW |
| DC.subject | 預測 | zh_TW |
| DC.subject | 深度學習 | zh_TW |
| DC.subject | 卷積神經網絡 | zh_TW |
| DC.subject | ITS | en_US |
| DC.subject | Traffic flow | en_US |
| DC.subject | Forecasting | en_US |
| DC.subject | Deep learning | en_US |
| DC.subject | Convolutional neural network | en_US |
| DC.title | Multi-headed CNN on Short-term Traffic Flow Forecasting in Taoyuan Urban Area | en_US |
| dc.language.iso | en_US | en_US |
| DC.type | 博碩士論文 | zh_TW |
| DC.type | thesis | en_US |
| DC.publisher | National Central University | en_US |