Study Of Combination Forecasting For Intercity Train Passenger Volume Based On HHT

The prediction of the passenger volume will be the main basis when the railway transport departments make passenger transport plans, design transport products and make marketing strategies. On the practical application of the forecasting, if the single-way prediction methods are combined properly, the result will be more accurate. In this thesis, a new forecasting method for intercity train passenger volume has been advocated, which combines Hilbert-Huang Transform(HHT) with Combination Forecasting method. By research on this approach, it can forecast train passenger volume effectively, improve the accuracy of the prediction results and solve the problem of instability by the single-way prediction methods. This approach can provide not only the passenger volume forecasting algorithm, but also the scientific basis for intelligent pre-Ticket, train plan design, revenue management and other services. The main contents of this thesis include:(1) In the study of the single-way prediction method, the thesis used fuzzy support vector to improve Support Vector Regression(SVR) method and optimize its parameter by Artificial Fish Swarm Algorithm(AFSA). Besides, the study expanded the application of the existing prediction methods. Numerical experiments show that the accuracy of the algorithm can improve the prediction results effectively.(2) In this thesis, time series model based on HHT and Back Propagation(BP) neural network model based on HHT are established. While using genetic algorithm to optimize the parameter of BP neural network, the purpose of optimizing predictive model can be achieved by scientific selecting parameters. Compared with the traditional forecasting model, these two models for numerical experiments can predict the intercity train passenger volume and improve the accuracy and relevance of the prediction.(3) By Empirical Mode Decomposition(EMD) method, the intercity train passenger volume frequency data are separated and by grouping prediction irrelevant interference factors are excluded. In these ways, the prediction accuracy of the results can be further improved. The data processing experimentally confirmed HHT can effectively improve the prediction accuracy. HHT can make the length correlation data become shorter correlation data and increase the correlation between the forecasting result and the passenger volume historical data effectively.(4) After the establishment of single-way forecasting model, combination forecasting model based on HHT is established to make the input information become more comprehensive. By linear and nonlinear combination forecasting principles, the weight of individual model is determined. This combination forecasting model is used to forecast the intercity train passenger volume and the final results show that the model not only has good reliability and stability, but also improves the accuracy of the prediction. Therefore, this model has a good reference in practical application.