模擬微觀混合車流下幹道號誌連鎖策略之研究

Abstract

台灣地區每日尖峰或交通需求量高時道路上阻塞情形嚴重，汽機車數量龐大且夾雜行駛於道路上造成車流的混亂與交通安全的疑慮，造成行車秩序之混亂。過去國外對於車流問題之研究著重於純小客車流，惟台灣地區之車流行為屬於混合車流，是故車流模式需加入機車行為，而早期以純小客車流所設計的號誌系統亦需重新規劃。 本研究參考過去混合車流之相關研究，採用二維座標法及行為門檻模式建立車流推進模式，機車行為則以動態車間距、改良式行為門檻模式模擬汽機車於道路上之實際行為，以系統內車輛總旅行時間最小延滯，與幹道方向最大續進量兩目標作為系統號誌時制之績效值，並同時考慮幹道號誌連鎖系統之三因子：週期、幹道綠燈時比、時差，以窮舉法方式，遞增數值以求得最佳化的系統績效值。此外，為驗證模擬車流的正確性及號誌設計之有效性，本研究以台中市中區單向道系統作為實證範圍，進行車流狀況調查，包含車輛進入速率、期望速率、車輛組成、車道產生比例、轉向比等，並以韋布斯特(webster)延滯公式加以計算在不同的週期、時比及時差下之系統績效值，同時亦考慮不同流量水準及機汽車混合比的狀況下之建議最佳時制，供相關單位進行幹道號誌規劃時設定使用。 根據本研究驗證結果，在設計有機車停等區的狀況下，因可讓機車先行紓解，故可稍微提升車輛的續進率並使系統績效微量增加；幹道綠燈時比為影響系統績效值最重要因素，車流屬高流量水準的狀態下，需增加幹道的綠燈時比並週期、減少時差，促使車輛的停等延滯時間減少，若屬低流量水準，則建議可以短週期時間、長時差以增加績效值。機車混合比高時，短週期的設計將利於車輛的續進及延誤時間減少，將可促使幹道系統的車輛運行績效最佳。

Traffic congestion has long term been one of the major control problems during rush hours in many cities of Taiwan. Field investigation has shown that the variability of driving behaviors of autos and motorcycles along roadways may increase disturbance and instability in traffic flow. Unfortunately, most of the literature focused only on the operations of auto traffic. The traffic conditions and control environment concerned are quite different from those in most areas in Taiwan. Thus, coordinating the existing traffic studies with the native research on mixed traffic flow may help realistically reflect and capture more details of the traffic world in Taiwan. With the inspiration from paper review, this study classifies the overall traffic behaviors along an urban arterial to three, i.e., intersection flow behaviors, inter-node link flow behaviors, and approach flow behaviors. Driving behaviors are modeled by referring to the behavioral threshold model and the two-dimensional coordinates model. Coordinated signal timing plans including cycle length, offset and phasing splits are designed using the exhausted searching algorithm in order to obtain the best performance value. To verify the accuracy of model development and the effectiveness of signal control, the characteristics of mixed traffic flow, such as approaching speeds, desired speeds, traffic composition, lane distribution, and turning ratio on a two-node one-way arterial in Taichung City were investigated and statistically analyzed. Initial signal cycle lengths were estimated by using the Webster's formula based on a variety of entering volumes and mixed flow ratios. System performance is evaluated by concurrently considering the indices of system delay minimization and throughput maximization. The simulation results showed that the vehicle progression rate and system performance may be increased when an exclusive stopping area for motorcycles is installed. The phasing split design is found to be a key factor that may affecting the system control performance. When at high entering volume, stop delay time may be reduced with increased cycle lengths and phasing splits, and decreased offsets. When at low entering volume, a combination of shorter cycles with longer offsets may lead to a better system performance. Also, as the proportion of motorcycles is high, a short cycle may have a higher advantage in increasing throughput and decreasing delays.