Prediction of HV Factor in Expressways Using A Short-Term VDS Data

Abstract

Generally expressway imply arterial highway that provides users for high levels of safety and efficiency with full control of access only for through traffic in the expressway. However, most of expressways are often experiencing congestion problems caused by highly mixed rate of heavy vehicles and concentration of vehicles depending on the time period and the travel direction in the expressways. So, it is strongly raised to identify the traffic system by traffic characteristic studies in the basic expressway segments.

The purpose of this study was to collect the 14-day vehicle detection system(VDS) data(volume, speed, headway, occupancy, and density, etc.) at about 180 or more stations of major expressways(Gyeongbu and Namhae expressways in Korea), analyze the traffic flow characteristic data, especially hourly volume factor(K) and estimate (, , , and on the basis of average daily traffic for 1, 3, 5 and 7 days, respectively) calculated in the basic expressway segments, construct the proper regression models between the hourly volume factor and estimate , and finally select the most appropriate model between the calculated and predicted hourly volume factors in the basic expressway segments.

From the traffic characteristic analyses, the analyses of hourly volume factor(K) and estimate , and the development and verification of model in the basic expressway segments, the following conclusions were drawn:

ⅰ) Traffic characteristics appeared to show a considerable difference in the direction of the basic expressway segments. So, it was needed to establish the expressway traffic management system based on the directional traffic characteristics for improving the efficiency of expressway.

ⅱ) Hourly volume factor(K) in the direction of expressways appeared to have a highly positive correlation with estimate (j=1, 3, 5, and 7) for a short-term period. So, it was needed to examine the relationship between hourly volume factor(K) and the estimate for each direction of the expressways.

ⅲ) The highest hourly proportions of K in expressways appeared to show the rural and urban traffic flow characteristics. So, it was needed to classify these hourly proportions of K for in-depth analysis into

0.06≤K＜0.07, 0.07≤K＜0.08, 0.08≤K＜0.09, 0.09≤K＜0.10, 0.10≤K＜0.11, 0.11≤K＜0.12, 0.12≤K＜0.13, 0.13≤K＜0.14, and 0.14≤K＜0.15.

ⅳ) Linear, quadratic, cubic and power models appeared to have a highly correlation coefficient(r) between hourly volume factor(K) and estimate (j=1, 3, 5, 7) for each range of interval. So, it was needed to select the proper model in predicting the hourly volume factor(K) with a high explanatory power().

ⅴ) Power model appeared to be very appropriate in predicting the hourly volume factor(K) by estimate with a high explanatory power() and validity for all ranges of interval. So, it was needed to verify the power model between the hourly volume factor(K) and estimate for a short-term period.

ⅵ) Model verification results appeared to show the high correlation coefficients(r) in the power model with estimate and fall inside Accept region for all ranges of interval. So, it was needed to determine the power model as the most appropriate one for predicting K in expressways showing the rural and urban traffic flow characteristics.