航空学报 > 2021, Vol. 42 Issue (8): 525800-525800   doi: 10.7527/S1000-6893.2021.25800

基于LSTM的飞行数据挖掘模型构建方法

王志刚, 王业光, 杨宁, 米禹丰, 曲晓雷   

  1. 航空工业沈阳飞机设计研究所, 沈阳 110035
  • 收稿日期:2021-04-15 修回日期:2021-05-08 发布日期:2021-05-26
  • 通讯作者: 米禹丰 E-mail:dzfq_138957@sohu.com
  • 基金资助:
    国家级项目

Construction method of flight data mining model based on LSTM

WANG Zhigang, WANG Yeguang, YANG Ning, MI Yufeng, QU Xiaolei   

  1. AVIC Shenyang Aircraft Design and Research Institute, Shenyang 110035, China
  • Received:2021-04-15 Revised:2021-05-08 Published:2021-05-26
  • Supported by:
    National Project

摘要: 提出了一种基于LSTM (Long Short Time Memory)模型的飞行历史数据挖掘模型的构建方法,此模型可以将飞行数据中有价值的目标数据自动提取出来。首先,通过滑动窗口法获得待检测数据;然后,将预先做好的训练样本数据输入到所构造的LSTM模型中进行训练,得到数据挖掘模型;最后,将待检测数据导入到训练好的LSTM模型中进行模式识别,将目标数据片段挖掘出来。结果表明,基于LSTM模型的飞行数据挖掘模型构建方法通用化程度高,可用于挖掘不同类型的目标数据,且识别率高,具有很高的工程应用价值。

关键词: LSTM, 飞行历史数据, 数据挖掘, 模式识别, 模型构建

Abstract: Based on Long Short Time Memory (LSTM) model, a method of constructing the flight history data mining model, which can automatically extract the valuable target data in flight data is proposed. First, to obtain the data to be detected by sliding window method. Then, to train the pre-made training sample data into the constructed LSTM model to get the data mining model. And finally, to model the pattern recognition of the detected data into the trained LSTM model. The results show that this method of constructing flight data mining model has high generalization degree, and can be used to mine different kinds of target data. The result also show that the recognition rate has high engineering application value.

Key words: LSTM, flight history data, data mining, pattern recognition, model construction

中图分类号: