Research On Bus Passenger Flow Analysis And Forecasting Based On Granular Computing

Bus is an important part of city public transportation,and bus passenger flow is the basis of the data of bus operation,scheduling and route planning.Predicting bus passenger flow and its trends can help transport management planning and managed the vehicle more scientific.Thus,fully grasped the change rule of passenger flow and to predict the future passenger flow has great practical significance.With the rapid development of science and technology,the traditional bus data processing system cannot calculated and analysis the historical data intelligently,and cannot meet the actual needs of scheduling.Due to it cannot grasp the future trend of passenger flow lead to the phenomenon that bus route planning are not unreasonable,resources cannot be fully utilized,part of the bus line waste of resources seriously,passenger satisfaction is getting lower and lower.In order to effectively solve the problem of inaccurate traffic prediction,this paper studies on analysis and forecasting the traffic flow.Based on the real passenger flow data of Guangdong Province,this paper analyzed the factors which influenced the passenger such as passenger flow periodicity,weather and holiday factors,then got the conclusion that passenger flow has characteristic of variation,uncertainty and complexity.Granular computing has unique advantages in the data processing of bus passenger data,and can compensate for the limitations of the traditional support vector machine in the processing of large sample data.Thus,this paper put forward the theory of granular computing combined with support vector machine to predict bus traffic flow.When using granular computing theory to predict bus traffic flow,this paper proposed combined rough set with support vector machine prediction model,using rough sets attribute reduction algorithm to deleting redundant attributes,and then using support vector machine regression model to forecast the passenger flow when assigned the time period.Simulation result shows that this model can reduce the sample complexity and improve the prediction accuracy when compared with other prediction models.In the aspect of predicting the trend of passenger flow,this paper also put forward the methods of using fuzzy granulation algorithm divided sample data to into coarse particles and the used support vector machine to predict the range and trend of the traffic flow over a period of time,through the simulation results shows that the model can predict the traffic flow in the future accurately.