基于列车运行安全的青藏铁路大风预测优化模型与算法  

Qinghai-Tibet Railway Gale Forecasting Optimization Model and Algorithm Based on Train Running Safety

作　　者：刘辉[1] 潘迪夫[1] 李燕飞[1] 

机构地区：[1]中南大学轨道交通安全教育部重点实验室,长沙410075

出　　处：《武汉理工大学学报（交通科学与工程版）》2008年第6期986-989,共4页Journal of Wuhan University of Technology（Transportation Science & Engineering）

基　　金：国家科技支撑计划项目资助(批准号:200603805200);铁道部科技研究开发计划项目资助(批准号:2006G040-A)

摘　　要：为减轻青藏铁路恶劣大风天气对列车行车安全的影响、对沿线风速进行准确地预测预报,运用时间序列法对格尔木-拉萨段16号测风站实测风速建立时序预测模型,并进行多步预测仿真计算.为提高时序预测模型精度,通过改进时间序列法建模流程,引进卡尔曼滤波智能算法,提出了2种适合于不同预测步长和精度的优化算法.预测实例表明:优化算法将时序模型的超前1步预测平均相对误差从4.89%降低为2.51%,超前5步预测平均相对误差从9.77%降低为5.62%,并明显改善了时序模型的预测延时现象.Gale along Qinghai-Tibet railway affects trains＇ operation safety. In order to give a high-precision prediction models for wind speed series along the line, which can help guide train operation dispatching, and reduce wind hazard of trains, using time series method, the paper establishes forecasting models for datum measured from 16th wind station along the Golmud-Lhasa line, and makes multistep forecasting simulation. To improve the forecasting accuracy by time series model, authors modify the model steps of time series method, introduce the Kalman filer method, and propose two optimization algorithms that are suitable for different prediction steps and precision. Experimental results indicate. optimization algorithms reduce MRE （Mean Relative Error） of one step forecasting from 4.89% to 2.51%, five step forecasting from 9.77% to 5.62%, and improve time-delay problem in forecasting.

关 键 词：时间序列 卡尔曼滤波 风速预测 优化模型 青藏铁路 

分 类 号：U298.1[交通运输工程—交通运输规划与管理] U283.4[交通运输工程—道路与铁道工程]