摘要
考虑到传统方法在监测航空器机翼积冰风险时受到积冰参数的影响,导致航空器机翼积冰风险监测频率偏低的问题,提出了基于C4.5决策树的航空器机翼积冰风险监测仿真研究。根据航空器机翼积冰风险值,对航空器的机翼故障区域分类,预测航空器机翼积冰风险,通过计算每一个风险节点相关属性信息增益,形成C4.55决策树的根部和各个节点相关的属性分裂树,完成对航空器机翼积冰风险数据决策树的构建。将原始的航空器机翼积冰风险数据完成多平面分割,过拟合处理航空器机翼积冰风险数据,实现航空器机翼积冰风险监测。仿真结果表明:提出的方法不会受到积冰参数的影响,可有效控制航空器机翼积冰风险,具有更好的风险监测能力。
In traditional methods,the frequency of monitoring icing risks of aircraft wing is low.Therefore,a simulation study on icing risk monitoring of aircraft wing was proposed based on C4.5 decision tree.According to icing risk values,the fault regions of aircraft wing were classified,and then the icing risk was predicted.By calculating relevant attribute information gain of each risk node,the root of C4.5decision tree and the attribute splitting tree related to all nodes were formed to construct a decision tree of icing risk data.Moreover,original icing risk data were used to split multiple planes.Furthermore,the icing risk data of aircraft wing were over-fitted.Finally,the monitoring on icing risks of aircraft wing was achieved.Simulation results show that the proposed method is not affected by icing parameters and can effectively control the icing risk of aircraft wings,with better risk monitoring ability.
作者
吴涛
王占海
陈奇
熊辉
WU Tao;WANG Zhan-hai;CHEN Qi;XIONG Hui(China Academy of Civil Aviation Science and Technology,Aviation Accident Investigation Center of CAAC,Beijing 100028,China;Air Traffic Control Center,Civil Aviation Flight University of China,Guanghan Sichuan 618307,China)
出处
《计算机仿真》
北大核心
2023年第7期44-48,共5页
Computer Simulation
基金
民航局安全能力建设项目(ASSA201909,ASSA201912,ASSA201915)。
关键词
决策树
积冰风险
航空器机翼
风险预测
过拟合处理
积冰参数
Decision tree
Risk of ice accretion
Aircraft wing
Risk prediction
Over fitting
Icing parameter
