%0 Journal Article
%T 基于时间序列的编组站中时排异重塑预测方法<br>Marshalling yard transit time rejection recast forecast method based on time series analysis
%A 张晓栋
%A 董宝田
%A 陈光伟&lt
%A br&gt
%A ZHANG Xiaodong
%A DONG Baotian
%A CHEN Guangwei
%J 铁道科学与工程学报
%D 2017
%X 针对现行铁路编组站车辆中时预测方法不能满足实际运输组织需求，提出一种基于时间序列的中时预测方法。根据编组站数据特点，建立中时序列，利用聚类分析方法检测显隐性异常，设计中时下界异常判定条件排除正常中时值，采用邻域插值方法对异常数据重塑，结合ARMA过程分析重塑中时序列并建模，对模型进行参数估计和启发式算法定阶，最终预测编组站短期中时。研究结果表明：该预测方法能够准确发现异常数据，序列重塑后效果较好，预测结果与实际生产情况符合度较高，可以很好地应用在对编组站中时的预测中，有利于编组站合理预测和分析运输生产活动，提高运输组织管理水平。<br>In practical production, according to the mismatch of marshalling yard transit time forecast method and practical production needs, the authors proposed a new transit time forecast method based on time series. On the basis of marshalling yard data characteristics, a transit time series were simulated. The dominant-recessive anomaly in the series was detected by the cluster analysis method, and the normal data was excluded by the transit time low-border determinant condition. Then the anomaly data was recast by the neighborhood interpolation method. The recast transit time series was analyzed and modeled by the ARMA procedure. The model parameters were estimated and the model order was decided by the heuristic algorithm. Finally, the transit time was forecasted in a short-term. By experiment calculation, the anomaly data was detected accurately and the recast marshalling yard transit time series has a better form. The forecast result coincided with the measurement data, which is better than the original data for the forecast. The marshalling yard transit time forecast method facilitates the practical railway production and the improvement of the level of railway transportation management
%K 铁路运输
%K 中时预测
%K 聚类算法
%K 时间序列
%K ARMA模型
%K 球形簇&lt
%K br&gt
%K railway transportation
%K transit time forecast
%K clustering algorithm
%K time series
%K ARMA modes
%K spherical cluster
%U http://www.jrse.cn/paper/paperView.aspx?id=paper_316530