Estimation approach for average link travel time based on fuzzy C-mean method

In order to availably utilize taxi GPS data with low frequency and precisely estimate the average link travel time (ALTT), an estimation approach of ALTT and Improved Map-matching algorithm was proposed. After analyzing upstream and downstream intersections of link particularly, more accurate estimates the moment of a vehicle through the intersection. Using the time difference of upstream and downstream intersections as single vehicle link travel time (SLTT). We generate patterns for the high, middle and low speed level by calculating the SLTT using FCM. ALTT calculated by means of a certain weight using the center value of each cluster after confirming of cluster membership of SLTT data. Computation result shows that the relative tolerance is less between the measure value and computation value of ALTT.