Neural Networks Based Real-Time Transit Passenger Volume Prediction

摘要

Real-time transit passenger volume prediction is the basis of establishing an intelligent public transportation dispatching system, and of great significance to improving the level of service of public transportation systems. In light of the characteristics of transit passenger flow, this paper suggests a solution for real-time transit passenger volume prediction based on neural network, with the input variables being the passenger flow volume, forecasting date, time and weather, and the output variable being the forecasting value of real-time passenger flow. In addition to presenting the structure and calculation method of the neural network model, this paper analyzes ways in collecting and managing related data information, examines the merits and weaknesses of such model and finally points out the orientation of the research efforts.