基于混合智能算法的铁路运量预测研究

发布时间：2018-04-15 05:22

本文选题：铁路运量预测 + 混合智能算法　；参考：《兰州交通大学》2012年硕士论文

【摘要】：近年来随着我国经济持续快速增长,铁路事业得到了跨越式发展,在建的铁路项目很多,还有很多的项目即将陆续开工。运量预测是铁路建设项目前期工作的核心内容之一,运量预测的水平和质量将直接影响到项目决策的科学性,而运量预测水平和质量在很大程度上取决于采用的预测方法。因此分析铁路运量的预测方法,研究如何提高预测的准确性和科学性是十分必要的。本文旨在将人工神经网络、改进粒子群优化算法组合的混合智能算法引入铁路运量预测领域中,并以预测精度为目标,研究上述人工智能算法在铁路运量预测中的应用。论文的主要内容如下：论文首先分析了铁路运量预测的重要性,总结了目前铁路运量预测的常用方法,并分析了各种方法的使用范围和优缺点。在此基础上提出分别以BP神经网络、灰色神经网络与改进的粒子群优化算法融合的两种混合智能算法进行铁路运量预测的思想。其次分别叙述了BP神经网络、灰色神经网络(GNN)的结构及其学习算法过程。接着研究了标准粒子群优化算法(PSO),在此基础上提出一种基于非线性权重变化的改进粒子群优化算法(IPSO),以四个标准测试函数验证了其有效性。然后通过采用IPSO算法优化BP神经网络连接权值,建立基于IPSO-BP神经网络的铁路运量预测模型,并以某阶段全国铁路客运量和货运量实际情况为研究背景进行了仿真研究与对比分析,仿真研究结果验证了此方法是有效可行的。另外进一步研究了铁路运量在信息不足条件下使用GNN建立预测模型的方法。在此基础上采用IPSO算法对GNN的白化参数进行优化改进了其不足,保证了预测结果的精度;同时利用灰色关联分析法,计算了铁路运量和其影响因素间的关联度,通过选用最主要的几个关联因素,建立了基于IPSO-GNN的铁路运量预测模型,并进行了仿真研究与对比分析,仿真研究结果表明了此方法是有效可行的。最后,对全文进行了概括性总结,提出了一些需要完善的研究工作,并指出了在铁路运量预测方面理论和应用上有待进一步研究的问题。
[Abstract]:In recent years, with the sustained and rapid economic growth in our country, the railway industry has been developed by leaps and bounds. There are many railway projects under construction, and many projects will be started soon.The prediction of transportation volume is one of the core contents in the early stage of railway construction project. The level and quality of volume prediction will directly affect the scientific nature of project decision, and the level and quality of volume prediction depend on the forecasting method to a great extent.Therefore, it is necessary to analyze the forecasting method of railway transportation volume and to study how to improve the accuracy and scientificalness of the forecast.The purpose of this paper is to introduce the hybrid intelligent algorithm based on artificial neural network and improved particle swarm optimization algorithm into the field of railway traffic forecasting, and to study the application of the above artificial intelligence algorithm in railway traffic forecasting with the aim of forecasting accuracy.The main contents of the thesis are as follows:This paper first analyzes the importance of railway volume prediction, summarizes the current common methods of railway volume forecasting, and analyzes the application scope, advantages and disadvantages of various methods.On this basis, two hybrid intelligent algorithms, BP neural network, grey neural network and improved particle swarm optimization algorithm, are proposed to predict railway traffic volume.Secondly, the structure and learning algorithm of BP neural network and grey neural network are described respectively.Then the standard particle swarm optimization (PSO) algorithm is studied and an improved PSO algorithm based on nonlinear weight variation is proposed. The effectiveness of PSO algorithm is verified by four standard test functions.Then, by using IPSO algorithm to optimize the weight of BP neural network, the prediction model of railway transportation capacity based on IPSO-BP neural network is established.Based on the actual situation of railway passenger and freight volume in a certain period, the simulation research and comparative analysis are carried out. The simulation results show that this method is effective and feasible.In addition, the method of using GNN to build prediction model of railway transportation volume under the condition of lack of information is further studied.On this basis, the whitening parameters of GNN are optimized and improved by IPSO algorithm to ensure the accuracy of prediction results, and the correlation degree between railway traffic volume and its influencing factors is calculated by using grey relational analysis.By selecting the most important related factors, this paper establishes the railway transportation forecasting model based on IPSO-GNN, and carries on the simulation research and the contrast analysis. The simulation results show that this method is effective and feasible.At last, the paper summarizes the whole paper, puts forward some research work which need to be perfected, and points out some problems that need further research in theory and application of railway transportation volume prediction.
【学位授予单位】：兰州交通大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：U294.13

【相似文献】