研究报告
韩博,姚婷玮,王立婕.基于航班运行规律的典型机场NOx浓度临近预测方法研究[J].环境科学学报,2022,42(3):362-372
基于航班运行规律的典型机场NOx浓度临近预测方法研究
- Research on NOx concentration prediction method of typical airport based on operation rules of flight
- 基金项目：国家自然科学基金项目（No.U1933110）；天津市教委科研计划项目（No.2018KJ248）
- 作者
- 单位
- 韩博
- 中国民航大学交通科学与工程学院， 天津 300300;中国民航环境与可持续发展研究中心， 天津 300300
- 姚婷玮
- 中国民航大学空中交通管理学院， 天津 300300
- 王立婕
- 中国民航大学空中交通管理学院， 天津 300300
- 摘要：针对天津机场区域考虑航班运行影响，应用广义加性模型（Generalized Additive Model， GAM）建立污染物浓度预测模型，对因子间的共线性问题和交互作用进行改进，得到最优的NOx浓度预测模型. 选取天津机场区域2019年11月—2020年3月环境、气象及航班数据，建立改进的GAM. 结果显示：①改进的GAM预测效果优，可以更加准确地预测浓度峰值及变化趋势；②样本量会影响模型选择的因子数量及模型性能；③改进模型adj-R2为0.940，实测和预测NOx浓度的相关系数为0.975，预测效果好；④航班活动对机场区域污染物浓度影响较大. 改进的GAM考虑污染物浓度与影响因子之间的复杂非线性关系及影响因子之间交互作用对污染物浓度变化的影响，使模型精度进一步提升. GAM对污染物浓度的准确预测可为机场区域污染防治提供依据.
- Abstract：The generalized model was used to establish the improved pollutant concentration prediction model. Meanwhile， concurvity and interaction among variables were modified to obtain the optimal pollutant concentration prediction model under the impact of flight operation. The data of Tianjin Binhai International Airport from November 2019 to March 2020 was selected to apply the model. The results showed that，①the modified generalized additive model has better prediction effect， and the modified model could predict the peak value and the trend more accurately. ②Sample size affected the number of factors and model performance③Furthermore， the adjusted R2 of the revised model was 0.940， and the correlation coefficient between the measured NOx concentration and the predicted NOx concentration was 0.975. ④Flight activities had a great impact on the NOx concentration in the airport area. The study showed that the improved generalization additive model considered the complex nonlinear relationship between pollutant concentration and influence factor and the influence factor interaction effect on pollutant concentration change， which further improved the accuracy of the model. The prediction of pollutant concentration by generalized additive model could provide basis for pollution control.
摘要点击次数： 209 全文下载次数： 259
