Study on Multidimensional Urban Public Transport Scheduling Method Based on Data Mining and Passenger Flow Prediction

Based on bus IC card data and bus GPS data, effective passenger flow information and vehicle operation information can be obtained through data processing, data analysis and mining, and corresponding predictions can be made to provide decision support for bus scheduling. For the data mining of IC card, this paper respectively puts forward the model of passenger flow time period division based on IC card and the passenger flow prediction model based on BP neural network. For the data mining of GPS bus, this paper studies the route matching of bus GPS, the analysis of the operation characteristics of bus in the process of operation, and the prediction of bus running time based on BP neural network. In terms of static scheduling, based on the minimum passenger waiting cost, the minimum crowding degree and the minimum operating cost of the bus company, the compilation model of the bus schedule is established and solved according to the genetic algorithm. For dynamic scheduling, this paper respectively studies the vehicle scheduling form under abnormal events, as well as site scheduling and inter-site scheduling in real-time scheduling to reduce the occurrence of crossovers and large intervals and improve the service level of the bus.