摘要：AbstractUrban traffic congestion has different typical characteristics under the influence of different conditions, such as different day of week, holiday and weather etc. It is necessary to set up the relationships between traffic congestion patterns and those influencing factors, when we conducting macroscopic analysis on the causes of traffic congestion. Based on Traffic Performance Index (TPI), a dynamic macroscopic index showing the whole area congestion intensity developed in 2007, typical congestion patterns are identified by using clustering method. A comparative analysis is conducted on setting rules for different clustering indexes. TPI pattern curves are derived and verified under combinations of date, transportation demand management policy, holiday, weather condition and etc., according to the actual traffic operational status. The analysis and verification results show that the method used in this paper is effective and feasible. TPI patterns indicate that traffic congestion has inherent characteristics which are primary and essential for transportation managers. This paper lays the foundation for traffic congestion prediction and early warning and proactive alleviation of traffic congestions.
关键词：Traffic congestion;Traffic Performance index (TPI);Congestion pattern curve;Cluster method;Data mining
