铁路沿线风速超前多步高精度预测方法研究

项目名称： 铁路沿线风速超前多步高精度预测方法研究

项目编号： No.51308553

项目类型： 青年科学基金项目

立项/批准年度： 2014

项目学科： 建筑科学

项目作者： 刘辉

作者单位： 中南大学

项目金额： 25万元

中文摘要： 强风是影响铁路运输安全的主要气象灾害之一。实现铁路沿线风速预测是铁路运营部门在恶劣强风环境下防范事故、科学决策和安全指挥调度的有效手段。针对目前风速预测所面临的超前步数有限、自学习能力差和求解复杂等问题，拟引入多种智能控制优化算法开展风速统计-智能、多智能混合预测研究，核心内容涵盖：① 运用信息闭环反馈技术对经典时序预测模型进行优化以提高其信息追踪融合能力；② 通过对小波/小波包分析法、模式经验分解法与优化后的闭环时序模型进行二次混合建模，建立不同组合的风速混合预测模型；③ 运用小波/小波包分析法、遗传算法等对经典神经网络模型进行优化以提高其对非平稳风速的全局搜索与高分辨识别能力，建立混合模型；④ 比较分析上述所建的风速预测模型体系，确定适应不同预测场合的铁路风速混合预测方法；⑤ 基于算法成果开发相应的风速智能预测系统及成套软件。

中文关键词： 铁路安全；风速预测；混合建模；超前多步；高精度

英文摘要： Strong-wind is one of major meteorological disasters which directly affect the safety of railway transportation. Realizing real-time high-precision forecasting for wind speed from certain important zones along strong-wind railway is an effective way to avoid the strong-wind caused accidents, and to provide a scientific guidance for railway departments. Aim at the disadvantages of limited ahead steps, poor self-learning capacity and hard computation in railway wind signal prediction, in this study a hybrid forecasting method based on several statistical and intelligent theories will be done. The main research contents are demonstrated as follows: ① To improve the non-stationary tracking performance of the traditional time series forecasting model, a modified way using information feedback idea will be presented; ② To establish different kinds of hybrid statistical-intelligent models, a new research using the upper modified feedback time series algorithm and some latest signal decomposition algorithms (including wavelet/wavelet packet/empirical mode decomposition) will be made; ③ To build different kinds of hybrid multi-intelligent forecasting models, a new study using genetic algorithm, artificial neural networks and some signal decomposition algorithms (including wavelet/wavelet packet/empirical mode decompositi

英文关键词： Railway security；Wind speed prediction；Hybrid modeling；Multiple steps；High accuracy