交通运输系统工程与信息

交通运输系统工程与信息 ›› 2015, Vol. 15 ›› Issue (1): 143-149.

• 系统工程理论与方法 • 上一篇    下一篇

采用支持向量机回归的航班延误预测研究

罗赟骞*1,2,3，陈志杰1,2，汤锦辉1,2，朱永文1,2   

  1. 1. 中国人民解放军95899部队，北京100085；2. 国家空域技术重点实验室，北京100085； 3. 中国人民解放军95865部队，北京100085
  • 收稿日期:2014-07-10 修回日期:2014-08-26 出版日期:2015-02-25 发布日期:2016-02-25
  • 作者简介:罗赟骞（1981-），男，四川名山人，工程师，博士.
  • 基金资助:

    国家重大科技专项（2013ZX03001028）；国家科技支撑计划（2011BAH24B10）

Flight Delay Prediction Using Support Vector Machine Regression

LUO Yun-qian1,2,3，CHEN Zhi-jie1,2，TANG Jin-hui 1,2，ZHU Yong-wen1,2   

  1. 1. The army 95899 of PLA, Beijing 100085, China; 2. National Key Laboratory of Airspace Technology, Beijing 100085, China; 3. The army 95865 of PLA, Beijing 100085, China
  • Received:2014-07-10 Revised:2014-08-26 Online:2015-02-25 Published:2016-02-25

PDF

摘要：

针对航班延误难以预测的问题，采用支持向量机回归方法建立航班到港延误预测模型.首先，采用相空间重构理论计算到港延误的延迟时间、嵌入维数和最大 Lyapunov 指数，发现到港延误时间序列存在混沌特性；将航班到港延误时间序列进行相空间重构，并结合执飞该航班的航空器在上游机场的离港延误构建模型的输入向量；其次，将粒子群算法、差分进化算法和遗传算法进行比较，用于选择最优的模型参数，实验表明，差分进化算法能够以较高概率获得最优的预测模型；最后，比较该模型、单一因素预测模型和相关向量机预测模型的航班延误预测性能.结果表明，该模型的预测性能明显优于另外两种模型，能够有效预测航班延误.

关键词: 航空运输, 航班延误预测, 支持向量机回归, 航班延误, 相空间重构, 差分进化算法

Abstract:

To solve the problem that the flight delay is difficult to predict, the support vector machine regression method is used to establish the flight arrival delay prediction model. First, the phase space reconstruction theory is used to calculate the flight arrival delay’s the delay time, embedded dimension and maximum Lyapunov exponent, and the chaotic characteristics of the flight arrival delay time sequence is found. The phase space of the flight arrival delay time sequence is reconstructed and combined with the departure delay of the upstream airport’s flight using the same aircraft to build the input variable vector of the prediction model. Second, for selecting the optimal model parameters, the particle swarm algorithm, differential evolution algorithm and genetic algorithm are compared, the experiment shows that differential evolution algorithm can get the optimal prediction model with a higher probability. Last, the prediction performance of the model, the single factor prediction model and the relevance vector machine prediction model are compared. The results show that the prediction performance of the model is much better than the other two, the model can effectively predict flight delays.

Key words: air transportation, flight delay prediction, support vector machine regression, flight delay, phase space reconstruction, differential evolution algorithm

中图分类号: