Adaptive Airport Taxi Dispatch Algorithm Based on PCA-WNN

Abstract

As a part of public transportation, taxi plays a very important role of passengers' travel from the airport to downtown. In order to exploring an efficient taxi dispatch mechanism and saving passenger waiting times, we develop an adaptive airport taxi dispatch system based on principal components analysis wavelet neural network (PCA-WNN). A series of new online short term time forecasting techniques are used to capture the relationship between taxi supply and demand. Then we proposed an adaptive feedback-based taxi dispatch algorithm for the effective response to the non-stationarity of taxi service under a changing environment. By using the real data of Beijing Capital International Airport within twenty weeks, our experiment demonstrated that this algorithm can predict accurately of the taxi service data and greatly improve the efficiency of taxi management.