通用航空训练飞行发动机数据异常检测初探  

Preliminary Study of Anomaly Detection of Engines of Training Flight in General Aviation

作　　者：王翔 WANG Xiang(Civil Aviation Flight University of China,Guanghan 618307,China)

机构地区：[1]中国民用航空飞行学院,广汉618307

出　　处：《内燃机与配件》2022年第3期97-99,共3页Internal Combustion Engine & Parts

基　　金：中国民航飞行学院青年基金“基于深度学习的航空发动机故障识别技术”(编号:Q2019-055)。

摘　　要：大型运输机发动机的健康管理研究较为广泛,相对于运输航空,针对通用航空领域以训练飞行为主的小型教练机发动机的异常检测技术还不够成熟。训练飞行具有飞行模式固定,起降频次较高,信息数据结构简单的特点,更适合引入深度学习对其进行建模分析。本文归纳了当前主流的几种深度异常检测模型,从原理、计算复杂度和优缺点三个角度进行分析。为通用航空训练飞行的教练机发动机的异常检测研究提供可行的研究思路。The research on health management of large transport aircraft engine is more extensive.Compared with transportation aviation,the anomaly detection technology of trainer aircraft engine focusing on training flight in the field of general aviation is not mature enough.Training flight has the characteristics of fixed flight mode,high takeoff and landing frequency and simple information data structure.It is more suitable to introduce deep learning for modeling and analysis.This paper summarizes several mainstream deep anomaly detection models,and analyzes them from the perspectives of principles,computational complexity,advantages and disadvantages.It provides a feasible research idea for the anomaly detection of the engines of training flight in general aviation.

关 键 词：异常检测 训练飞行 深度学习 

分 类 号：V216.7[航空宇航科学与技术—航空宇航推进理论与工程]