基于贝叶斯网络的航班保障服务时间动态估计

发布时间：2018-04-07 17:41

本文选题：航班保障服务　切入点：机器学习　出处：《计算机应用》2017年01期

【摘要】：针对航班保障服务时间估计的问题,考虑到航班保障服务流程的特殊性、复杂性以及影响因素的不确定性,提出了一种基于贝叶斯网络(BN)的航班保障服务时间估计模型。该模型把航空领域的专家知识与历史数据的机器学习相结合,使用贝叶斯网络的增量学习特性动态地调整BN模型,使其适应新的变化,进而不断更新航班保障服务时间的估计值。使用国内某大型枢纽机场信息系统内提取的数据,通过期望最大化(EM)方法对模型进行训练,得到了测试结果。实验结果分析与模型评价表明,所提方法能有效估计航班保障服务时间且具有较高的准确度。敏感性分析表明,航班到达时段的航班密度对航班保障服务时间影响最强。
[Abstract]:Considering the particularity, complexity and uncertainty of the flight support service process, a model of flight support service time estimation based on Bayesian network (BN) is proposed.This model combines expert knowledge in aviation field with machine learning of historical data, and dynamically adjusts BN model to adapt to new changes by using incremental learning characteristics of Bayesian network.Furthermore, the estimated time of flight guarantee service is constantly updated.Using the data extracted from the information system of a large hub in China, the model is trained by the method of expectation maximization (EMM), and the test results are obtained.The experimental results and model evaluation show that the proposed method can effectively estimate the flight guarantee service time and has a high accuracy.The sensitivity analysis shows that the flight density of the flight arrival time has the strongest influence on the flight guarantee service time.
【作者单位】：中国民航大学电子信息与自动化学院;中国民航局第二研究所信息技术分公司;
【基金】：国家自然科学基金委员会-中国民用航空局联合研究基金资助项目(U1533203) 中央高校基本科研业务费基金资助项目(3122014P003)~~
【分类号】：V35;TP18

【相似文献】