民用飞机进近着陆阶段灾难事故类型预测  

Prediction of Catastrophic Accident Types of Civil Aircraft at Approach and Landing Phases

作　　者：郭媛媛[1] 孙有朝[1] 李龙彪[1] 胡宇群[1] 

机构地区：[1]南京航空航天大学民航学院,江苏南京210016

出　　处：《航空计算技术》2016年第4期31-34,共4页Aeronautical Computing Technique

基　　金：国家自然科学基金委员会与中国民用航空局联合项目资助(U1333119);国防基础科研计划项目资助(JCKY2013605B002);工信部民机专项资助(MJ-F-2011-33);上海民用飞机健康监控工程技术研究中心开放课题基金项目资助(GCZX-2015-05);国防科工局技术基础科研项目资助(Z052013B003);江苏省自然科学青年基金项目资助(BK20140813)

摘　　要：为减少民用飞机进近着陆阶段发生灾难事故,给出了1985年至2013年灾难性事故样本,获得了3种主要事故类型和14种关键致因。提出了一种灾难事故类型的预测方法,建立了事故类型预测流程,样本事故致因作为输入层,样本事故类型作为输出层,基于BP神经网络和Elman神经网络进行多次训练和仿真。结果表明,预测的事故类型与实际情况基本吻合。针对事故致因有效判断民用飞机在进近着陆阶段潜在事故,给出纠正措施,保障航空运行安全。In order to reduce disaster accidents of civil aircraft at approach and landing phases,the catastrophic accidents samples have been listed at these specific phases from 1985 to 2013. Three main types of catastrophic accidents and its 14 key reasons are also given from the samples. Moreover,a prediction method of catastrophic accident types has been proposed and the prediction process of these accidents types are given at approach and landing phases. With accident reason as input layer,and accident type as output layer,the train and simulation have been carried out many times based on BP neural network and Elman neural network. The results show that the actual values closely match the predicted values with an acceptable level of error. The corrective measures can be taken based on the effective judgment of accidents that might happen at approach and landing phases.

关 键 词：民用飞机 进近着陆 事故类型 BP神经网络 ELMAN神经网络 

分 类 号：V328.5[航空宇航科学与技术—人机与环境工程]