基于CNN-LSTM混合模型的民航非计划事件分析方法OACSTPCD
Analysis Method of Civil Aviation Unplanned Events Based on CNN-LSTM Hybrid Model
安全是民航业的核心主题,非计划事件是辨识安全隐患、改善航空安全的重要信息来源.非计划事件的非结构化和数量庞大等特性使得人工分析变得困难且效率低下.为提高非计划事件的分析效率和精度,论文提出了一种基于CNN-LSTM的混合深度神经网络模型,用于民航非计划事件的自动化分析.并与SVM、CNN、LSTM等模型进行比较,在航空公司的事件日志样本数据集上进行训练并做出事件分类结果的判断.实验结果表明,所提出的CNN-LSTM混合模型具有最高的分类准确率,对于不平衡数据样本,具有最稳定的分类性能.
Safety is the core theme of the civil aviation industry,and unplanned events are an important source of information for identifying safety hazards and improving aviation safety.The unstructured and large number of unplanned events make manual analysis difficult and inefficient.In order to improve the analysis efficiency and accuracy of unplanned events,this paper proposes a hybrid deep neural network model based on CNN-LSTM,which is used for the automated analysis of civil aviation unplanned events.It is compared with SVM,CNN and LSTM models,trained on the airline event log sample data set,and judged the event classification results.The experimental results show that the proposed CNN-LSTM hybrid model has the highest classification accu-racy,and has the most stable classification performance for unbalanced data samples.
王捷;周迪;左洪福;陆扬
南京航空航天大学民航学院 南京 211106东南大学电气工程学院 南京 210018
计算机与自动化
深度学习民航安全文本分析卷积神经网络长短时记忆神经网络
deep learningcivil aviation safetytext analysisconvolutional neural networklong short-term memory
《计算机与数字工程》 2024 (006)
1714-1720 / 7
国家自然科学基金项目(编号:U1933202);民航大NSF重点基金项目(编号:U1733201)资助.
评论