A methodology to estimate overall travel time from individual travel time measurements within a time window is presented. To better handle data with complex outlier generation mechanisms, fuzzy clustering techniques have been used to represent relationships between individual travel time data collected within a measuring time window. The data set is considered to be a fuzzy set to which each data point belongs at some degrees of membership. This allows transitions from the main body of data to extreme data points to be treated in a smooth and fuzzy fashion. Two algorithms have been developed based on ‘point’ and ‘line’ fuzzy cluster prototypes. Iterative procedures have been developed to calculate the fuzzy cluster centre and the fuzzy line. A novel estimation method based on time projection of a fuzzy line has been proposed. The method has the advantage of being robust by using a wide time window and the timeliness by employing time projection in resolving the most recent travel time estimation. Unlike deterministic approaches where hard thresholds need to be specified in order to exclude outliers, the proposed methods estimate travel times using all available data and, thus, can be applied in a wide variety of scenarios without fine tuning of the threshold.
Estimation of travel time using fuzzy clustering method
IET Intelligent Transport Systems ; 3 , 1 ; 77-86
2009-03-01
10 pages
Article (Journal)
Electronic Resource
English
On-Line Travel Time Estimation Using Fuzzy Neural Network
British Library Conference Proceedings | 1995
|TRAVEL TIME ESTIMATION SYSTEM AND TRAVEL TIME ESTIMATION METHOD
European Patent Office | 2015
|An Improved Fuzzy Trajectory Clustering Method for Exploring Urban Travel Patterns
DOAJ | 2021
|Transportation Research Record | 2021
|