With the increasing problems occurred frequently in traffic jams and accidents, increasing attention has been paid for the forecast for the rail transit passenger flow. Meanwhile, the ability to correctly predict the normality transit passenger flow plays a vital role for the analysis of large-scale events, weather and other unusual factors' impacts on passengers. In this paper, based on PSO-SMO (Particle Swarm Optimization-Sequential Minimal Optimization), a rail transit normality passenger flow forecasting model can be presented as follows. Firstly, a specific model can be built in use of the SMO algorithm. And then, combination of the model built above, the method of PSO-CV (Cross Validation) is introduced to optimize parameters. Finally, the model with optimal parameters extracts the training sample characteristics and the prediction passenger flow can be output. The results show that more than 80% of the data points' prediction relative error are less than 10%, certificating the validity of the model.
展开▼
