神经网络于拥挤指针之研究

拥挤是国内外用来描述路况最通俗的代名词，主要在于简单易懂；目前以车速间距发布拥挤等级的方式常会发生与用路人主观之行车拥挤感知经验不符的现象。文中以路段固定侦测器之实时交通参数、CCTV信息画面，结合类神经网络理论，探究群体用路人于号志化干道上之拥挤感知。以台15线为例，进行主观拥挤指针之模式建构与评估。

Analysis of Congestion Index Using Neural Network

Congestion is the most common word to describe the traffic condition because of its easy understanding and simplicity. However, it is usually inconsistent with the congestion perception of driver＇s experience when the currenttraffic congestion index is interpreted by different levels in terms of speed range. This paper integrates the real-time road detector data, CCTV image, and Neural Network to explore the true subjective congestion perception for road drivers.The arterials route of No. 15 is selected to test the performance of proposed model.