چكيده به لاتين
Nowadays, the ever increasing growth of population and the increase of immigration to urban areas and metropolises, led to increase the number of intra city trips and consequently road traffics. This causes impose of heavy expenses, including time and pollution to the society. Short-term traffic planning requires the proper perception of the future conditions of routes. Accordingly, the accurate estimation based on previous observations can be suggested to properly represent the future conditions. Travel time information not only helps the users to save time, but also is considered as a fundamental concept for evaluating the road system operation for better management. In this research, the travel time perdition based on time series and historic data is conducted for one of the main axes of Iran (Tehran-Qom), one of the most heavily traveled and congested roadways in Iran. The database was collected from the Bluetooth traffic detector. Bluetooth is a telecommunications industry specification that defines the manner in which mobile phones, computers, personal digital assistants, car radios, and other digital devices can be easily interconnected using short-range wireless communications. In principle, the Bluetooth traffic monitoring system calculates travel times by matching public Bluetooth wireless network IDs at successive detection stations. The time difference of the ID matches provides a measure of travel time and space mean speed based on the distance between the successive stations. The test site was about 118 kilometers and covered two sensors. 90 day period (3 months) was used from the historical database, 70% to do modeling and forecasted next 30% day values. Modeling was done in Two categories, for weekdays during the period from 7 am to 8 pm, as well as daily (for each day of the week) in the 24-hour period. After identifying the model type, possible models were estimated. Autoregressive moving average (ARMA) was implemented. After analyzing the model's parameters, the best model was selected based on the Akaic criteria (AIC) and Beesin Schwartz (SBC). Finally, after checking the lack of correlation, a desirable model is proposed to predict travel time. The results of each model from the case studies are investigated and reported. All in all, the results models can be used for decision making and traffic management accurately.
