機車專用車道車流模式建立之研究

Abstract

國內一般道路上常常會見到汽機車於車流中產生相互干擾及衝突的情形，因而影響到整體道路容量、車流順暢與行車安全性。為了解決長期汽機車混流所帶來的交通問題，陸續已有相關研究投入於設置機車專用道以將汽機車進行分流之研究，提出透過交通工程的設計與方法，能有效提昇車流行車秩序與安全性。然而，由於設置機車專用道將會形成單純機車車流的出現，故針對機車專用車道之機車流的駕駛行為特性分析與模式構建有其必要性。國外對於機車車流研究相當缺乏，反觀國內雖有不少關於機車車流的相關研究，然其多以混合車流中汽機車為研究對象，針對單純機車車流之微觀機車駕駛行為等基礎研究尚不夠完整，因此有必要加以探究，以建立更完整的微觀機車車流模式。 本研究目的為構建與驗證機車專用車道之機車車流模式，針對機車專用車道之直線路段為研究範圍，並以跟車理論、變換車道理論與模糊理論為理論基礎，採用文獻評析法、統計迴歸分析、攝影調查法及人工智慧演算法等作為本研究方法。本研究所構建之車流模式包括跟車模式及超車模式，其中跟車模式主要係以實際行駛於機車專用車道之駕駛者跟車行為的車流資料進行調查，並分析其主要影響因素，再透過模糊推論結合類神經與遺傳演算法等二種人工智慧方式應用於機車跟車模式構建，且進一步比較兩者之優劣；而超車模式則透過超車偏向角之特性分析，進而構建超車偏向角模式，其後再建立超車行為決策流程及準則，以完整描述機車專用車道之機車車流，作為機車專用車道機車車流研究之基礎。因此，本研究結果將可供改善交通工程與設計的參考依據，以及控制與管理所需之機車車流模式。

In Taiwan, most traffic flow is mixed automobiles and motorcycles flow. This situation influences the road capacity, traffic flow order and safety. To solve this problem, some reviewed literature emphasizes setting up the motorcycle exclusive lane by traffic engineering method, could improve traffic order and safety effectively. However, motorcycle traffic flow study is quite scarce abroad. In Taiwan, though there are a lot of studies of motorcycle traffic flow, they are aimed to mixed traffic flow rather than motorcycles exclusively. As a result, the microscopic motorcycle traffic flow study is incomplete. This study is attempted to structure and verify microscopic motorcycle traffic flow models on motorcycle exclusive lane by carrying on the investigation and analysis with the real traffic flow materials on the section road. This study is based on car-following theory, lane-changing theory and fuzzy theory. Motorcycle car following and overtaking model were built with literature review, statistics analytic approach, video investigation method and artificial intelligence algorithms etc. Car following model was built by the fuzzy inference approach, combining artificial neural network and genetic algorithm, and further compare the quality of them. Overtaking model was built by analyzing the characteristics of the angle of overtaking, and then setting up overtaking decision procedure and criterions for describing the motorcycle traffic flow of the motorcycle exclusive lane completely. As a result, this study results could serve as the basis of improving the motorcycle traffic planning and design, and control.