基于CNN-GRU-Attention的民机重着陆预测模型  

Hard landing prediction model for civil aircraft based on CNN-GRU-Attention

作　　者：吴翔鑫 余汇[2] 任艳丽[1] WU Xiangxin;YU Hui;REN Yanli(School of Communication and Information Engineering,Shanghai University,Shanghai 200444,China;Shanghai Aircraft Design and Research Institute,Shanghai 201210,China)

机构地区：[1]上海大学通信与信息工程学院,上海200444 [2]上海飞机设计研究院,上海201210

出　　处：《电子设计工程》2024年第13期41-45,共5页Electronic Design Engineering

摘　　要：重着陆事件是民用飞机着陆阶段最容易出现的事故之一,对其进行预测对于提升飞行安全具有重大意义。因此提出了一种基于卷积神经网络(Convolution Neural Network,CNN)和门控循环单元(Gated Recurrent Unit,GRU)并融合了注意力机制的重着陆预测模型,以飞机实际运行过程中采集的快速存取记录器(Quick Access Recorder,QAR)数据作为数据集。对QAR数据进行预处理并通过轮载信号确定数据集区间;通过Spearman相关系数对与飞机着陆相关的参数进行相关性分析,从中提取了24个特征参数作为预测模型输入,以重着陆判定指标垂直加速度作为输出,建立基于CNN-GRU-Attention的重着陆预测模型。实验结果表明,所提模型与其他预测模型相比具有更好的预测效果。The hard landing event is one of the most common accidents during the landing phase of civil aircraft,and predicting hard landing events is of great significance to improve flight safety.Therefore,this paper proposes a hard landing prediction model based on Convolution Neural Network(CNN)and Gated Recurrent Unit(GRU)and incorporating attention mechanism.The Quick Access Recorder(QAR)data collected during the actual operation of the aircraft is used as the dataset.Preprocess the QAR data and the data set interval is determined by the Weight on Wheel signal.The correlation analysis of the parameters related to aircraft landing was performed by Spearman correlation coefficient,from which 24 characteristic parameters were extracted as the input of the prediction model,and the vertical acceleration,the index of hard landing determination,was used as the output to establish a CNN-GRU-Attention based hard landing prediction model.The experimental results show that the proposed model has a better prediction effect compared with other prediction model.

关 键 词：民用飞机 重着陆预测 卷积神经网络 注意力机制 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]