基于SVR-随机森林模型的民机落地剩油预测研究  

Fuel Reserve Prediction of Civil Aircraft Landing Based on SvR and Random Forest Model

作　　者：林熙颢 韦冬冬 唐盛香 钱宇[2] Lin Xihao;Wei Dongdong;Tang Shengxiang;Qian Yu(Shanghai Aircraft Customer Service Co.,LTD.,Shanghai,201100,China;Civil Aviation Flight University of China,Guanghan,618307,Sichuan,China)

机构地区：[1]上海飞机客户服务有限公司,上海201100 [2]中国民用航空飞行学院,四川广汉618307

出　　处：《中国民航飞行学院学报》2023年第4期40-43,共4页Journal of Civil Aviation Flight University of China

基　　金：中国民用航空飞行学院科研创新团队(JG2022-22);中国商飞第四届科技周项目(COMAC-FKGS-2021-163)。

摘　　要：针对国产民机落地剩油的预测与单一算法预测精度较低的问题,本文提出了一种基于支持向量回归机(SVR)与随机森林(RF)算法组合的预测模型对国产民机落地剩油做预测。灰色关联度分析得到落地剩油关联度较高的因素,简化模型输入量;采用单一的SVR算法与RF算法进行落地剩油预测,利用倒数误差法将两个单一算法组合起来对落地剩油做预测。实例验证,单一的预测模型准确率为81.21%、83.91%;组合模型的准确率为93.2%,提高了落地剩油预测的精度,有利于飞机在安全飞行的前提下合理减少额外油,提高经济效益。Aiming at the problem of low prediction accuracy of domestic civil aircraft landing fuel reserve and single algorithm,a prediction model based on the combination of Support Vector Regression(SVR)and Random Forest(RF)algorithm is proposed to predict domestic civil aircraft landing fuel reserve.The factors with high correlation degree are obtained by grey correlation degree analysis,and the input quantity of the model is simplified.A single SVR algorithm and RF algorithm are used to predict fuel reserve,and the reciprocal error method is used to combine the two single algorithms to predict fuel reserve.The example shows that the accuracy of the single prediction model is 81.21%and 83.91%.The accuracy of the combined model is 93.2%,which improves the accuracy of landing fuel reserve prediction,and is conducive to reducing additional fuel reasonably and improving economical benefits underthe premise ofa safe flight.

关 键 词：落地剩油 SVR 随机森林 倒数误差法 组合预测 

分 类 号：V37[航空宇航科学与技术—航空宇航推进理论与工程]