| 引用本文: | 马子骥,唐涛,刘宏立,彭强,金滩.基于非等间距灰色模型和Elman神经网络的轨道质量预测[J].哈尔滨工业大学学报,2018,50(5):137.DOI:10.11918/j.issn.0367-6234.201707012 |
| MA Ziji,TANG Tao,LIU Hongli,PENG Qiang,JIN Tan.Forecasting of track quality based on unequal-interval grey model and Elman neural network[J].Journal of Harbin Institute of Technology,2018,50(5):137.DOI:10.11918/j.issn.0367-6234.201707012 |
|
| 摘要: |
| 轨道质量是影响行车安全的关键因素，合理预测轨道质量可以有效指导铁路工务部门进行轨道养护和维修.轨道质量指数(Track Quality Index, TQI)是综合评价单一区间段内轨道质量的参数.本文通过深入研究TQI的发展趋势，提出一种将非等间距灰色模型和遗传算法优化Elman神经网络相结合的预测方法.首先利用优化后的非等间距灰色模型GM(1，1)得到原始TQI序列的大致发展趋势，然后为了描述轨道质量发展中各因素之间复杂的函数关系，利用遗传算法优化后的Elman神经网络对初步预测结果进行残差校正，从而得到更为准确的TQI预测序列.新方法将轨道质量发展趋势中的随机波动成分纳入方法考虑范围，充分挖掘了历史数据的发展规律.利用沪昆线上行实测TQI数据对本文方法进行验证，实验结果表明：新方法对轨道质量发展中的随机波动趋势拟合效果较好；对于轨道质量预测，在利用非等间距灰色模型进行初步预测基础上，使用Elman神经网络进行残差校正，由此得到的预测结果在均方根误差、相对系数、决定系数等多个统计指标上均优于其他方法.
|
| 关键词: 轨道质量 非等间距 灰色模型 Elman神经网络 遗传算法 |
| DOI：10.11918/j.issn.0367-6234.201707012 |
| 分类号:U212.24+6 |
| 文献标识码:A |
| 基金项目:中央高校基本科研项目(1053214004)；国家自然科学基金资助项目(61771191)；湖南省科技计划重点项目(2015JC3053)；湖南省自然科学基金项目(2017JJ2052) |
|
| Forecasting of track quality based on unequal-interval grey model and Elman neural network |
|
MA Ziji,TANG Tao,LIU Hongli,PENG Qiang,JIN Tan
|
|
(College of Electrical and Information Engineering, Hunan University, Changsha 410082, China)
|
| Abstract: |
| Track quality is essential to track safety. A reasonable forecast of track quality is a good instructor for the department of railway maintenance when arranging track maintenance schedule. Track quality index (TQI) can evaluate track quality in a unique track interval. With research on the changing tendency of TQI, this paper proposes a forecasting method that combines unequal-interval Grey Model and Elman Neural Network. The Grey model GM (1, 1) is previously exploited to obtain an approximate forecast of original TQI series and then the residual error correction is corrected by using Elman Neural Network optimized by Genetic algorithm. The new method takes random fluctuation of changing tendency of TQI into consideration, thus the historical data can be treated more efficiently. The proposed method is demonstrated with practically measured data of Shanghai-Kunming Railway Line. The forecasting results show that, comparing to other forecasting methods, the method which uses Elman Network to correct residual error correction reaches higher forecasting accuracy at root mean square error, correlation coefficient and determination coefficient.
|
| Key words: track quality unequal-interval grey model Elman neural network genetic algorithm |
