The travel time being provided to drivers through the variable message signs and the car navigation equipment in real time, being called here the "provided travel time" is calculated with the travel time obtained by measuring the time differences between two points, being called here the "measured travel time". In the measured travel time, however, the times peculiar to vehicles to be measured such as times stopped by a convenience store and so forth are included. Therefore, there are cases that it is widely different from travel time corresponding to the traffic conditions in the section at that time. For this reason, it is indispensable to verify whether the measured travel time corresponds to the traffic conditions at that time. The verification is made generally by setting the upper and the lower threshold values of travel time for every record attribute, such as each month, day of week and time period, using record attributes of travel time measured in the past and checking whether they are within the threshold values. Consequently, there are no problems if the verification is made under the traffic conditions close to that of the past records, however, under the traffic conditions different from that of the past records, such as in the event of traffic congestion which was not kept in records, it becomes impossible to make an appropriate verification. To solve this problem, a new elimination method of abnormal values in the measured travel time by using the speed-concentration (U-K) model is described. In other words, using traffic volume at representative points within a section along with the measured travel time, a model of the relationship between the average travel speed of the section, which is calculated by dividing the section length by the average measured travel time, and the average concentration of the section, which is calculated by dividing the traffic volume by the average travel speed, is created to make a verification of the upper and the lower threshold values of the measured travel time. Making a study of applicability of this method to actual fields, it is verified that the method presented here is far superior to the conventional method. For the covering abstract see ITRD E134653.
Samenvatting