Study On Forecast Of Railway Traffic Volume Based On Hybrid Intelligent Algorithm

In recent years, with the sustained and rapid growth of China’s economy, Chinese railways have made a leaping development. There are many projects under construction, and some scheming projects are going to be carried out in succession. Railway traffic volume forecast is the core part in the preliminary work of railway construction projects. The forecast of traffic volume of passengers and freight can directly affect the scientificity of making decisions in projects, but the level and quality of forecast volume largely depends on the adopted forecast methods. Therefore, it is very necessary to analyze the forecast method and research how to improve its accuracy and scientificity.This paper is aimed to apply the hybrid intelligent algorithm to the forecasting field of traffic volume, which is composed of by the artificial neural network and the improved Particle Swarm Optimization (IPSO) algorithm. It is aimed at high forecast accuracy while researching the algorithm’s application in railway traffic volume forecast. The main contents of this article are as follows:This paper firstly introduces the importance of railway traffic volume forecast, summarizes the common forecast means, analyzes its application area and relative characters and proposes the idea based on two different hybrid intelligent algorithms, which are generated(formed) by the IPSO algorithm combining with BP network or Gray neural network (GNN).Secondly this paper discusses the structure and learning process of BP neural network and GNN.The next, this paper describes the standard Particle Swarm Optimization (PSO) algorithm, puts forward an IPSO algorithm based on the nonlinear weight variation and verifies its effectiveness by using four classical test functions.Then, a railway transport forecast model is established based on the IPSO-BP neural network, which optimizes BP neural network connection weights with IPSO algorithm. And using the actual situation to a certain stage of the national railway passenger and freight traffic volume as the research background, conducted a simulation research and comparative analysis, the simulation results show the IPSO-BP neural network forecast method is an effective and feasible tool for railway traffic volume forecast.In addition, a forecast model is also established by using GNN under the conditions of insufficient given information. With IPSO algorithm, optimizing the "whitening" parameters of GNN and improving its deficiencies ensure the precision of forecast outcomes. By using gray relation analysis method it calculates the correlation degree between railway transport volume and its influence factors. Then it chooses several important relative factors and establishes the forecast model based on IPSO-GNN, and conducted a simulation research and comparative analysis, the simulation results show the IPSO-GNN forecast method is an effective and feasible tool for railway traffic volume forecast.Finally, the paper gives a generalizing summarization, proposes some researches that needs improvement and points out future study of theory and application in this domain.