技术站列车晚点时间预测方法

摘要

Accurately forecasting the technology service station train delay time can not only be used for the station warning technology,but also set aside time for the station to arrange train route,as far as possible to ensure the goods to the deadline,so it is important to improve the abil ity of technology and market of railway freight station with guarantee.This paper analyses the train delay situation and the main influencing factors of technical station train delay.The GM (1,2) model is used to predict the train delay time and the Markov matrix is used for error correction in order to make sure the range in which the error lie,we can make the prediction in a short term be in the prediction interval by predicting and revising over and over again.In the meantime,we use RBF neural network(NN) to increase the training data by interpolating to predict the delay accurately and the average error is kept be-low 3％.Finally,the advantages and disadvantages of the two forecasting methods are analyzed and compared and we point out the application scope of the prediction method severally.%准确地对技术站整体列车到达晚点时间进行预测不仅可以为技术站调度预警,也可以为车站合理安排列车进路预留时间,从而尽可能保证货物运到期限,对技术站能力的提高和铁路货运市场的保证具有重要意义.本文分析技术站列车到达晚点情况和技术站列车晚点的主要影响因素.建立GM(1,2)模型对列车晚点时间进行预测,再利用马尔可夫矩阵进行误差修正,确定目标所处区间范围,往复预测并修正,使未来一段时间内的模糊预测结果准确地落在预测区间内;同时利用径向基神经网络(RBF)对数据进行插值扩充,从而对晚点时间进行精确预测,平均相对误差保持在3％以下.最后分析比较了两种预测方法的优劣并分别指出了预测方法的适用范围.