動態車流模擬與動態交通路網模式構建之研究

Abstract

動態車流模擬可用來預測道路上的車流變化情況，以作為交通管理者的決策分析工具，其亦可提供即時的交通路況資訊，為實現智慧型運輸系統所不可或缺的重要環節。 巨觀的動態車流模式乃是將車流行為視為一連續流體，而以雙曲型偏微分方程式來表示。然而，由於非線性偏微分方程式不易求解，通常沒有解析解的存在，且在不同的初始條件或不同的邊界條件下，將會導致不同的解，所以一般都需要透過數值方法來求取近似解。 過去相關的車流模擬研究幾乎都集中在討論單一路段中，單一車道上的車流行為，多車道上的車流行為尚未廣泛地被討論，且目前多車道車流行為的研究多以微觀的角度來進行。因此，本研究希望能夠以目前已經發展的單一車道巨觀動態車流模式為基礎，構建能夠描述多車道車流行為的動態車流模式，並以交通錄影資料來驗證模式構建時所做的行為假設。 除此之外，本研究尚將討論的對象延伸至交通路網的規模，藉由結合描述路段車流行為的巨觀動態車流模式與描述節點車流分配的巨觀路徑選擇模式構建一個完整的動態交通路網模式。

Dynamic traffic flow simulations can be used to forecast the variations of the traffic flow on a roadway section. Thus, it can be an analytic tool for making traffic control policies. Also, it can provide real-time traffic information, which is an important part in implementing the Intelligent Transportation System (ITS). Macroscopic dynamic traffic flow models regard the traffic flow as a continuous fluid, which is in the form of hyperbolic partial differential equation. However, the nonlinear partial differential equations are difficult to resolve, and there does not exist analytic solutions. Moreover, different solutions will be derived under distinct initial or boundary conditions. Therefore, it needs numerical schemes to obtain the approximations. Most of the studies of traffic flow simulation in the past are focused on single-lane roadway sections. There are few discussions about the traffic flow behavior on multi-lane roadway sections, and for the time being the studies of multi-lane roadway sections are mostly in the microscopic point of view. Thus, a multi-lane dynamic traffic flow model based on a proposed single-lane macroscopic dynamic flow model will be developed in this thesis, and the assumptions of traffic flow behavior will be verified by traffic video records. In addition, the discussion subject will be extend into traffic networks, and an overall traffic network model will be construct by combining the macroscopic dynamic traffic flow model describing the traffic flow propagation behavior on links and the macroscopic route choice model describing the traffic flow splitting behavior on nodes.