Passenger Flow Prediction Based on Particle Filter Optimization

The article aims at making scientific and accurate prediction of the current flow of passagers based on its characteristics, which is nonlinear and is influenced by various factors. Clustering is used to make classification of IC card data.Then ARMA prediction model is installed and the model parameter is worked out through matrix method and revised by the method of particle filtering. Resampling is also conducted through genetic optimization. The data of rail transit of Chongqing City of June, 2012 is used to make verification.The result shows that the prediction of optimize parameters through the method of particle filtering is close to the real values. The MAE is 0.8154,MAPE is 1.755,so this method in our article can make the prediction of passenger flow volume more accurately.