高速公路多車道車流模式之研究

Abstract

多車道路段由於可變換車道，其車輛行為與單車道時有所不同，故應以多車道車流模式描述，而過去車流研究所建立的多車道車流模式多為波動方程式而鮮少有探討多車道車流特性之曲線形狀。本研究透過台灣國道高速公路上之偵測器蒐集實際車流資料，將不同車道數之車流資料繪製成車流特性關係圖並觀察其曲線形狀，發現道路容量並不隨著車道數增加而等比例增加，但是道路壅塞時之流量增加比例卻大於車道數的增加比例。此外，本研究以相鄰車道間速率會相互影響的觀點，改寫Drake、Schofer和May的車流模式中之自由車由速率項，發展一高速公路多車道車流速率與密度之關係曲線方程式，並使用平均絕對誤差百分比(Mean absolutely percentage error, MAPE)做為模式績效指標。根據Lewis(1982)對此方法提供之建議尺度，本研究所建立之模式其MAPE在各種車道數下皆為10%以下，是高精準的預測模式。

This study presents a set of traffic flow models that account of operating effects among vehicles on multilane highways. It is known that the driver’s behavior and traffic characteristics on single-lane highways are fairly different from those on multilane ones. Many of the past studies have applied various wave equations to describe such traffic phenomena. This study explores the curve tendency, and their characteristics and associated parameters of each the traffic model that is proposed and calibrated from some freeway segments with different number of travel lanes. The actual traffic data are collected from a number of lane-based VDs along the National Taiwan Freeway. For two freeway segments with different lanes, the data analysis has shown that the rate of change in total capacity is not proportional to that in the number of lanes. When in congestion, however, the rate of increasing in congested flow is larger than that in the number of lanes. Plus, the free flow speed for a specific lane may be affected due to the side friction of traffic streams on the adjacent lanes. This study has modified the free-flow-speed term of the model proposed by Drake, Schofer, and May and developed a new speed-density relationship for multilane highways. The index of Mean Absolutely Percentage Error (MAPE) is used as the measure to check the fitness of each model to the associated actual data. Based on the scale suggested by Lewis in 1982, the MAPE value for each model with varied lanes is less than 10%. Hence, it is truly believed that the proposed traffic flow models can work with high accuracy.