| Summary: | 碩士 === 國立成功大學 === 交通管理學系碩博士班 === 98 === Advanced Traffic Management Systems (ATMS) and Advanced Traveler Information Systems (ATIS) aim to enhance the efficiency and safety of transportation systems. One of the key issues is to provide information based on accurate flow predictions. Travelers can thus make better decisions based on received real time traffic information.
Traffic information is estimated based on traffic flow models and these traffic flow models are usually calibrated based historical data. However, these models do not vary with respect to time, thus cannot reflect dynamic characteristics of real-time data.
This research aim to construct dynamic flow models based on observed real-time traffic data, such as flow and occupancy. This Modified Greenshield’s model is deployed to describe the speed-density relationship for both freeways and urban streets and the dynamic flow models are established with transfer functions to provide the linkage between models and real-time data. The mixed traffic flows in Taiwan are primarily composed of passenger cars and motorcycles. The area percentage of passenger car equivalents (PCE) is used to provide accurate estimation of the impact of motorcycles in the mixed traffic flows.
The actual flow data from vehicle detectors are used in the validation process. Two indices, the Mean Absolute Percent Error (MAPE) and the Root Mean Square Error (RMSE), are used to measure the error between the estimated and the actual speed data. The results show that both static and dynamic models can describe traffic flow reasonably. However, the dynamic models can reflect the variation of real-time data thus can predict flow characteristics more accurately with respect to real-time data.
|
|---|
