巷道減速墊設置準則之研究

Abstract

近年來巷道交通之理論研究及改善工程，有逐漸增多的趨勢，而國內交通寧靜設施應用最廣的即屬減速墊之設置，但在此過程中卻缺乏一套客觀完整的設置程序及準則。另外，相較於歐美等已開發國家，為數眾多的機車占台灣都市運具相當大的比例，因此本研究希望針對國內巷道系統及運具比例之交通特性，嘗試將汽機車車流做各別分析，分別建立巷道評估模式與巷道車速預測模式，從中建立一套減速墊之設置準則。 應用模糊理論中模糊隸屬函數之建立巷道風險評估模式，主要目的乃將行人對巷道車速的真實感受程度，透過問卷設計反映出來，分成極有威脅、有威脅、普通、安全、極安全等五等級。透過評估模式的建立，可以清楚劃分出行人在多快的汽機車車速下感覺受到威脅，可作為巷道風險之衡量指標。當巷道風險超過標準時，即表示該巷道有設置交通寧靜設施之需要。 巷道車速預測模式乃應用神經網路預測之功能，由巷道中歸納出影響車輛行駛速度之因素，如巷道幾何條件、車輛位於巷道之相對位置、減速墊位置、減速墊高度等，預測出車輛行駛於巷道之車速。模式建構過程包括網路訓練、網路測試及績效評估等程序。研究結果模式之預測能力績效良好。 藉由本研究構建出兩模式之結合，可以建立一套完整的巷道預測及評估方法。本研究亦透過實例分析，對目標巷道做完整的評估流程及方案建議，期望供決策單位作為參考依據，完備現行之巷道交通改善工程。

In recent years, more and more traffic calming projects are implemented to improve the poor traffic condition of local roads. Among various kind of traffic calming measurements, speed pad in the most in common use. But there are nearly none studies about developing a method for establishing application guidelines for installing speed pads local streets. On the other hand, motorcycle which having a high ratio in the transportation system in Taiwan is not a main transportation mode in overseas countries and very few studies were focused in motorcycle traffic. Based on this understanding, this study tries to distinguish the motorcycle traffic from automobile traffic by building separately risk estimation model and speed prediction model. By using the Fuzzy Theory, real pedestrian experience can be reflected form very threatened to very safe as the risk estimation model. Questionnaires are used in this study for data collection in the model-building process. Hence, we can tell from the model in what speed of which transportation mode is too much risk for a local road. Therefore, speed pads must be implemented once the level of traffic safety is unacceptable. With the use of Neural Network method in building the speed prediction model, vehicle speed will be predicted according to several variables such as geometric conditions of a local road, the position of the vehicle in the road, types of traffic calming measures within the road, and the traffic or pedestrian flow within the local road. The whole model-building process with parameter calibration, model verification and model validation has been completed in this study. Combining the two models, a complete method for estimating the local roads’ safety level and the effectiveness of implemented traffic calming measures is accomplished.