|摘要: ||由於交叉路口的車輛疏解特性對交叉路口的容量評估影響甚劇，而疏解率(departure rate)的應用常換算成飽和流率，進而推算出理論容量值。因此，車輛疏解特性因素，在整個道路口容量與分析中佔了相當重要角色。最近公路容量研究報告中發現飽和流率一直無法在觀察值中達到收斂的極值，近而對過去求算容量的方式產生諸多的疑點。從這些的觀察中，或許回歸到疏解特性的研究，重新檢驗研究的方式。|
號誌化交叉路口車流行為當中主要可分為直行、左轉、與右轉，本研究則以探討直行部份分析。而本研究希望採取兩種研究方向來進行調查以及剖析，一是對於混合車流(mixed traffic flow)的觀察方式採取，我們以每個車隊中前後兩輛車作為一組跟車樣本，並將前後兩輛車之車種，將其分類成小車-小車、大車-小車、小車-大車以及大車-大車等四組樣本。之後，了解這四組樣本之跟車疏解特性，並進行觀察不同跟車車種間疏解間距之差異性，以作為後續模式構建之依據。二是了解混合車流之疏解特性，以分析車隊結構之穩定位置求得，作為模式構建之依據。因此，我們將車隊劃分成兩個階段，一為不穩定階段，主要由小車疏解間距及大車效應(heavy vehicles effects)兩個函數，共同預測此階段之車輛疏解時間；二則為穩定階段，分別以平均疏解間距及微觀跟車間距函數預測此階段之車輛疏解時間。之後，我們以組合此兩階段模式，以建構兩種不同之車輛疏解時間預測模式，預期得到最佳效果。最後，我們以兩種不同型態之交叉路口，驗證疏解時間模式之合理性及可行性，都可以得到不錯之結果，其統計MAPE指標值均在15%以內，屬於優良預測。
Because of platoon dispersion characters at intersections impacting seriously on junction capacity, departure rates are often applied to change into saturated flow rates calculating theoretical capacity. Therefore, it played a quite important role in all the road capacity’s analyses. Recently, the research articles of the road capacity found out saturated flow rates converged on the observed data, and there are many problems on the way to calculate capacity in the past. From these findings, maybe it returned the study of departure characters, and renewed to examine the method of the research.
The behavior of traffic flow at the intersection mainly separated into straight, left, right, and we only discussed straight traffic flow. However, we hope to apply two research’s aspects to investigate and analyze. Firstly, how the observation methods applied in the mixed traffic flow, we chose forward-backward two vehicles of each platoon as one set of car-following sample, and according to two vehicles’ type, we classified four sets of vehicle samples as small-small, heavy-small, small-heavy, and heavy-heavy. Then, we realized the car-following departure characters of these samples, and began to observe the differences of each departure headways types. Secondly, after understanding the departure characters in the mixed traffic flow, we analyzed the platoon structure to get the stable position.
In addition, we divide platoon into two states, one is unstable state, there are major small-car departure headways and heavy-vehicles effects’ functions forecasting the vehicle departure time of the state; the other is stable state, we used average departure headway and microscopic car-following headway to predict the vehicle departure time of the state. Then, we combine these two state’s models to construct two different vehicle departure time models.
At last, we investigated two different type intersections to validate the rationality and feasibility of our models. We can get some best results, and the statistical MAPE indexes are among 15%，belonging to excellent forecast.