Research On Urban Taxi Travel Demand Based On GPS Data

Urban transport system is the basic condition for economic and social development of a city. In recent years, along with the rapid pace of China’s urbanization, the size of the city is expanding and the urban population continues to grow up substantially, which leads to the increasing of residents travel demand. Featured by the timeliness, convenience, flexibility, speediness, taxi has been an important transport option for city residents which provide the "point to point" transport service.At present, cruising taxi is the main type of the taxi service in China cities. Due to the randomness and flexibility of passengers’ travel and the blindness of vacant taxis when looking for the passengers in the road-net, there is always a contradiction between the taxi service supply and the residents’ travel demand. Henceforth, by taxi travel demand prediction, vacant taxis can be guided to the areas which harbor more travel demand so that the contradiction between the taxi service supply and the residents’travel need can be eliminated.Based on the available literature, this paper summarized the present researches on the operation and management of taxis, the taxi GPS data applications and the traffic short-term prediction. Compared with the present research result, this paper focused on the travel demand estimating and predicting of taxis from the view of the city regional characteristics.Firstly, on the basis of taxi GPS data preprocessing, this paper stressed the taxi travel demand estimation algorithm based on the griding method. Through the three steps of griding, demand estimation and grid-merging, the paper estimated the scope of the study area total travel demand in a certain period of time.Secondly, by distinguishing the different land use type as the separate mode method, the research selected an office park, a commercial district, a transportation hub area, campus area and hospital area as five typical study zones to analyze the time-varying characteristics of travel demand in different regions which provides the basis for the taxi travel demand forecasting model’s establishment.Finally, taking the QiLu Software Park as the case study, this paper made the analysis of the relating factors’influence to the predicator variables. Based on the Kalman Filter model and Artificial Neural Network model with the genetic algorithm optimization, this paper presented a combined prediction model. In order to demonstrate the effectiveness of the combination prediction model, the paper compared the Kalman Filter model, GA-BP Neural Network model and the combination prediction mode in three diverse scenes.