基于EEMD-ELM 的航班运行风险混沌预测

针对航班运行风险可靠预测方案，以某航空公司2016-2018 年航班运行风险数据为基准，通过相空间重构，序列混沌特征的识别，构建基于极端学习机(ELM)的航班运行风险混沌短期预测模型，基于集成经验模态分解(EEMD)阈值降噪方法进行改进；最后，计算风险预测结果，分析不同方式下的预测精度. 结果表明：航班运行风险时间序列具有混沌特征， EEMD方法可抑制序列本征模态函数(IMF)的模态混叠现象；经由EEMD阈值降噪处理后，短期预测结果的修正平均绝对百分误差(MAPE)值显著下降. 证实本文航班运行风险预测方案可行且有效.

Chaotic Prediction of Flight Operation Risk Based on EEMD-ELM

To developa reliable prediction methodon flight operation risk, according to the flight operation risk data of a certain airline in 2016-2018, through the reconstruction of phase space reconstruction for the time series, and the identification of series' chaos characteristics, an Extreme Learning Machine (ELM) based chaotic shortterm prediction model for flight operation risk was constructed combining with an iterative prediction method. The model then is improved by Ensemble Empirical Mode Decomposition (EEMD) threshold de-noising method. Finally, the risk prediction results are calculated, and the prediction accuracy of different methods is compared. The results show that the series has chaos characteristics. EEMD method can suppress modal aliasing of the IMF sequences. After EEMD threshold de- noising, the revised MAPE for the short forecast results is significant reduced. It is confirmed that the operation risk prediction method in the paper is applicable and effective.