兩階段模糊邏輯控制之匝道儀控細胞自動機模擬

Two-stage Fuzzy Logic Control for Ramp Metering with Cellular Automaton Simulation

本文建構獨立型及整合型之兩階段模糊邏輯控制（fuzzy logic control,FLC）高速公路匝道儀控模式。其中，第一階段FLC之原理係依據即時偵測所得之交通資訊判斷主線之車流型態（分為自由流、輕微同步流、強烈同步流、大範圍擁擠流）及其隸屬度，獨立型及整合型模式均相同；第二階段FLC則在決定當時最適之儀控率，惟獨立型及整合型模式分別有不同之考量因素。獨立型模式考量第一階段FLC輸出之車流型態判斷隸屬度及匝道等候長度，整合型模式則進一步考量上游匝道之儀控率，以發揮上下游連鎖效果。為驗證模式並深入探索匝道儀控點附近車流之時空特性，本文亦發展細胞自動機（cellular automata,CA）模擬模式，並與定時儀控及ANCONA儀控方式（Kerner,2005）進行績效比較。經由一15.5公里之雙車道高速公路測試路段（涵蓋3處交流道）模擬結果發現，在不同交通量情境下，整合型模式比不實施匝道儀控可減少2.16％至6.66％之總旅行時間，且均優於其它儀控方式。進一步觀察匝道附近地區之車速時空變化發現，本文所建構之獨立型及整合型模式皆可提高平均車流速率，降低速率變化，且有效避免壅塞車流向上游擴散，具實用價值。在應用上，當交流道間距較大時（如郊區），建議可採獨立型匝道儀控模式；當交流道間距較小時（如都會區），則以整合型模式為佳。

This paper develops two-stage fuzzy logic control (FLC) models for metering an isolated freeway ramp and integrated freeway ramps with coordination between upstream and downstream metering rates. The first-stage FLC for both isolated and integrated ramp-metering models has the same rationale, which determines the mainline traffic phase (free-flow, light synchronized, heavy synchronized, or wide moving jam) based on the real-time traffic information detected. The core logic of the second-stage FLC is to determine the proper metering rate: for isolated ramp it is based on the on-ramp queue length and the traffic phase determined in the first stage; whereas for integrated ramps it is based on the on-ramp queue length, the traffic phase determined in the first stage, and the upstream ramp metering rate. Cellular automata (CA) traffic simulation is employed to validate the proposed two-stage FLC ramp metering models and to explore in-depth the spatiotemporal traffic patterns in the metering stretches under various metering strategies, including the proposed isolated and integrated FLC metering, no metering, pre-timed metering and ANCONA metering proposed by Kerner (2005). A 15.5-kilometer two-lane freeway mainline section composed of three interchanges is tested. The results show that our proposed integrated FLC ramp metering model has consistently outperformed the other metering strategies. Using the no metering strategy as the baseline, the integrated FLC ramp metering can curtail the total travel time by 2.16% to 6.66% under different scenarios of traffic volume tested. Moreover, the in-depth exploration on the spatiotemporal vehicular speed profiles indicates that both isolated and integrated FLC ramp metering can significantly increase overall average speed, reduce speed variation, and eliminate jam propagation to upstream. For practical applications, the isolated FLC ramp metering is recommended in rural areas where interchanges are normally sparsely located; in contrast, the integrated FLC ramp metering is recommended in urban areas where interchanges are oftentimes closely located.