Research On The Similarity Of Freeway Traffic Flow

This paper analyzes traffic behavior, and researches the repeatability and similarity of traffic flow data occures on day/week cycles on the basis of large amounts of traffic data, using statistical method or other mathematical methods.In this paper, research on similarity of traffic flow data could be transformed into time series similarity mining by using traffic flow time series, relating to many problems, such as the evaluation method of similarity and similar mode classification.This paper primarily focus on the following tasks: obtaining enormous data of freeway traffic flow in Shenzhen City, filtering the original data of transport flow statistically, and restoring the loss part by linear method; discussing the definition and using method of each kind of similarity criterion, including distance measure methods and qualitative approach. Analyzing the data with amount of traffic flow rate imformation of cities and comparing with the experiment results, by which the existing of similarity of the traffic flow can be improved; innovatively proposing the analytical method of wave band and successfully deviding the traffic flow data into varietal modes in one month by this may. One day's traffic flow pattern could be identified by comparing and analyzing the difference between this day's traffic flow data and every pattern's wave band quickly. In this paper, the existing of much traffic flow pattern is verified by analyzing the month-mean's pattern, the workdays' pattern and the vacations' pattern in details. Meanwhile, it can prove the repeatability and similarity of traffic flow data occures on day/week cycles indirectly.