2005 Fiscal Year Final Research Report Summary
Dynamical properties of traffic flow -Data analyses and simulation-
| Project/Area Number |
15607014
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
計算科学
|
|---|
| Research Institution | Saga University |
|---|
Principal Investigator |
TADAKI Shin-ichi Saga University, Computer and Network Center, Professor, 総合情報基盤センター, 教授 (00202169)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
KIKUCHI Macoto Osaka University, Cybermedia Center, Professor, サイバーメディアセンター, 教授 (50195210)
|
|---|
| Project Period (FY) |
2003 – 2005
|
| Keywords | 1 / f fluctuation / traffic flow / observed data / time sequence analysis / DFA |
|---|
| Research Abstract |
Researches on physical models of traffic flow and their simulations have been activated in viewpoints of dynamical phase transition and pattern formation. These researches have unveiled fundamental properties of traffic flow. Real traffic data, however, have not been studied better than modeling and simulations. The purpose of this research project is to study real traffic data observed Japanese expressways and to compare them with simulation results.
This research project treated the observed traffic data of Japanese expressways including Tomei expressway. The project has been succeeded to clarify phenomena, which have not been understood well. It has been pointed out that the flow in expressway shows 1/f fluctuation. Previous researches show that 1/f fluctuation can be observed two distinct regions, one or two hours and several days. Between these two time regions, there is daily periodic behavior. Weekly and monthly periodicity may be included too. Therefore it is not clear which time scale the 1/f fluctuation is included in.
This research employed the DFA (Detrended Fluctuation Analysis) method to analyze the time sequence of traffic flow. The method detracts trends and extracts long-range correlations included in the time sequence. The time sequence is divided into non-overlapping segments. The deviation from the trend in segments is analyzed. The dependence of the deviation on the segment length shows the long-range correlation. We applied the DFA method to the yearly data and obtained the clear result that the traffic flow contains the 1/f fluctuation. The 1/f fluctuation is shown not to depend on traffic jam.
|
