Method for High-way States Analysis Based on Clustering Algorithm

摘要

Since the traffic flows are complicated and unstable, there is no standard to classify the traffic flows around the management of traffic, which causes the obstacle to the managers. The purpose of this study is to use flow, velocity, occupancy as input parameters and build up a traffic state classification model based on clustering algorithm. Furthermore, based on the traffic flow theory, this study presents a new method to identify the initial center in clustering in order to avoid the traditional flaws and improve the efficiency in clustering algorithm. Finally, the study utilizes samples to validate the differences and improvement of modified K-means model and modified FCM model. The results prove that modified FCM model is more suitable for the need in traffic management. This model is able to give the exact definition of traffic states, which may discriminate congestion state of high-way and support management of traffic.