以模糊推論系統與細胞自動機方法探討混合車流環境下機車行進行為

Abstract

充分了解車輛之行進行為不但是道路規劃與設計之基礎，更是研擬控制與管理策略之參考依據，但傳統車流模式多為大型車或小客車而研究，鮮少以機車為主要考慮對象，特別是混合車流中的機車行進行為，因機車並無行駛特定寬度之車道特性，而常穿梭於車輛間，其行進方式與小客車並不全然相同，因此現有的車流模式並無法真正反映機車之行進行為。由於機車為國內非常普遍之交通工具，深入了解機車於道路上之行進行為模式乃相當重要之課題。 本研究之主要目的即在解析顯著影響機車行進行為之因素，並根據其特性構建可以描述機車行進行為之模式。根據相關特性分析結果，我們提出了機車跟車模式及非均質顆粒跳動模式來描述機車在混合車流中的行進行為。前者係根據顯著影響因素，分別構建了GM及模糊推論模式來描述機車在混合車流中之跟車行為，後者則藉由適當的細胞自動機(CA)法則來描述車輛在離散的時空環境中之車輛（含機車與小客車）間互動行為，在此系統中，每輛車都依循不同環境下所預設之控制規則移動，以產生不同時點上的速率與位置。 因此，本研究首先透過現場調查，獲取混合車流中機車及與旁車（含機車及小客車）間互動所呈現之二維座標行進軌跡，並整理出可衡量機車行進行為之相關變數，如機車本身之速率、加速率、機車與鄰車的側向間距、鄰車之速率、機車所在之橫向位置等因素。其中，採用顯著影響機車跟車行為之的因素建立機車一維行進模式－跟車模式，以描述機車加速率與影響因素之關係。其次，本研究應用CA法則以控制機車行進邏輯，發展一二維行進模式，而根據這些CA法則，以決定所有的車輛在每一時間之速率與位置。進一步，本研究藉由另一組現場調查資料來驗證所提CA模式之結果，此外，並利用此模式模擬在不同道路寬度與車種組成下之情況。最後提出結論與未來研究方向建議。

Understanding the vehicle moving behaviors provides the fundamental rationales for planning, designing, controlling and managing the road systems. However, conventional traffic flow models are developed mostly in depicting the moving behaviors of heavy vehicles (bus, truck) and light ones (car). Little is devoted to motorcycles’ moving in the mixed traffic context. Unlike heavy or light vehicles that normally move along within a specific longitudinal lane and sometimes change lanes for overtaking or turning, motorcycles do not move in a specified lane. Thus, conventional flow models may not satisfactorily elucidate the motorcycles’ moving behaviors. Because motorcycles are the most popular transportation mode in Taiwan as well as in some other Asian countries, it is important to gain better insights into the motorcycles’ moving behaviors, both from academic and practical perspectives. This study is to identify the significant factors that influence motorcycles’ moving characteristics and establish the motorcycle moving models. We propose motorcycle-following models and inhomogeneous particle-hopping models to describe motorcycles’ moving behaviors in mixed traffic with cars and motorcycles. The former models are established with GM and fuzzy-based models, the later models are established with appropriate cellular automaton (CA) rules in such a way that the speed of each vehicle (car or motorcycle) changes in discrete time steps as a consequence of its interactions with other vehicles. Each vehicle also follows certain pre-assigned rules that govern the positions of the vehicle over time and space, depending on various circumstances. A field observation is conducted to collect the two-dimensional trajectories and basic properties for motorcycle movements. The significant factors affecting motorcycle-following behaviors are used for constructing the one-dimensional moving models that properly describe the relationship between motorcycle acceleration rates and the related factors. Furthermore, this study develops two-dimensional models with CA rules governing the motorcycles’ moving logics. According to these CA rules, the instantaneous states of positions and speeds of all vehicles in any time frame can be determined. Our CA models are validated by another set of field data and then further applied to simulate various traffic compositions and lane widths. Finally, some suggestions for future research are presented.