在线咨询
中国工业与应用数学学会会刊
主管：中华人民共和国教育部
主办：西安交通大学
ISSN 1005-3085  CN 61-1269/O1
工程数学学报

工程数学学报 ›› 2023, Vol. 40 ›› Issue (2): 171-189.doi: 10.3969/j.issn.1005-3085.2023.02.001

• •    下一篇

基于CNN-BiLSTM网络模型的无人机飞行质量评价

罗  晶1,  高  永2,  梁葆华3,  刘军民1,  惠永昌1   

  1. 1. 西安交通大学数学与统计学院，西安 710049 
    2. 海军航空大学航空基础学院，烟台 264001 
    3. 中国飞行试验研究院，西安 710089
  • 收稿日期:2020-11-26 接受日期:2021-03-29 出版日期:2023-04-15 发布日期:2023-06-20
  • 通讯作者: 高 永 E-mail: gaoyongbh@sina.com
  • 基金资助:
    国家自然科学基金 (61877049)；科技创新计划 (201809164CX5JC6).

UAV Flight Quality Evaluation Based on CNN-BiLSTM Network Model

LUO Jing1,  GAO Yong2,  LIANG Baohua3,  LIU Junmin1,  HUI Yongchang1   

  1. 1. School of Mathematics and Statistics, Xi'an Jiaotong University, Xi'an 710049
    2. College of Aeronautical Basic, Naval Aviation University, Yantai 264001
    3. Chinese Flight Test Establishment, Xi'an 710089
  • Received:2020-11-26 Accepted:2021-03-29 Online:2023-04-15 Published:2023-06-20
  • Contact: Y. Gao. E-mail address: gaoyongbh@sina.com
  • Supported by:
    The National Natural Science Foundation of China (61877049); the Science and Technology Innovation Plan (201809164CX5JC6).

PDF (PC)

75

摘要：

为了更好地挖掘无人机飞行轨迹数据中蕴含的有效信息，准确客观地基于轨迹数据对无人机飞行质量进行评价，提出一种融合卷积神经网络 (CNN) 和双向 (Bi-directional) 长短期记忆 (LSTM) 神经网络的 CNN-BiLSTM 网络模型。首先，利用 CNN 网络和 BiLSTM 网络分别获取飞行轨迹数据的局部卷积特征和时间特征。然后，将两种特征送入特征融合层，使用融合后的特征进行分类并获得评分标签。针对六个数据集的数值实验表明，模型不仅取得了较好的分类效果，而且具有很好的泛化能力。

关键词: 无人机质量评价, 卷积神经网络, 双向长短期记忆网络

Abstract:

In order to better mine the effective information contained in the UAV flight path data and evaluate the UAV flight quality accurately and objectively based on the trajectory data, a CNN-BiLSTM network model which integrates convolutional neural network (CNN) and bi-directional long short-term memory (LSTM) neural network is proposed. Firstly, CNN and BiLSTM are applied to obtain the local convolution feature and time feature of flight path data respectively, and then the two features are sent to the feature fusion layer, and the fused features are used to classify and get the rating label. The numerical experiments on six datasets show that the model not only achieves good classification results, but also has good generalization ability.

Key words: UAV quality evaluation, convolution neural network, bi-directional long short-term memory network

中图分类号: