Research On Key Technologies Of Stop Planning And Traffic Flow Prediction In Intelligent Transportation Systems

With the rapid development of economy and the continuing progress of urbanization in China, the scale and population of cities expand quickly which have brought great pressure on urban traffic management. However, limited transportation infrastructures cause lots of difficulties, such as traffic congestion, accidents, air pollution and so on. Serious traffic problems have become an important factor to hinder urban development. Intelligent transportation systems (ITS) have emerged as an efficient way of improving the performance of transportation systems, reducing traffic congestion and air pollution, etc.Public transportation systems are important components in ITS. As the key functions, bus stop planning and short-term traffic flow forecasting are able to efficiently improve management and service of urban public transportation. Unfortunately, traditional bus stop planning methods mainly relied on subjective human surveys to capture people’s mobility patterns. Although these approaches are proved to be feasible in some sense, time and cost spent are quite substantial. Short-term traffic flow forecasting can effectively ease traffic congestion, but timely and accurate short-term traffic flow forecasting is still a technical difficulty. Meanwhile, the increasing number of traffic sensors deployed in facilities leads to the explosive growth of traffic data. We utilize the records of taxi GPS and bus transaction to research airport shuttle bus stop planning and short-term traffic flow forecasting service.To address the disadvantages of traditional bus stop planning approaches, we first propose a two-phase airport shuttle bus stop planning method. Our method aims at providing convenient public transit to the airport by identifying optimal airport shuttle bus stop. In the first phase, we divide the record data into two parts according to origin and destination of each record after data preprocessing. In the second phase, the k-means clustering algorithm is employed to identify candidate airport shuttle bus stops and then optimize these stops. Then, we propose an evolutionary method for the passenger flow prediction. In detail, based on the history record, we introduce ARIMA model to predict passenger flow volume of several periods. On the other hand, we update the future prediction based on the difference of the actual passenger flow volume and the predicted passenger flow volume. In order to help the manager in bus dispatch center make better dispatch scheme, we also proposed a method to determine a confidence interval of 1-ε. Finally, extensive experiments are conducted on a large-scale real-world taxi GPS data sets and bus transaction data sets to verify the practicality of our methods by using an In-memory database platform, HANA.