摘要
监测飞控系统状态参数是保证无人机飞行安全的重要手段。针对无人机飞控系统的组成特点和飞行控制律,设计并构建了基于长短期记忆网络(Long ShortTermMemoryNetwork,LSTM)的飞控系统状态监控模型。首先,利用无人机历史飞参数据训练模型,建立输入飞参数据与状态参数的回归映射关系;然后,利用训练好的网络模型,实时预测飞控系统的状态参数,通过对比实测值与预测值之间的差异,实现飞控系统的状态监控。选取无人机飞参数据进行实验,基于LSTM的算法比反向传播神经网络(BPNN)、支持向量机(SVM)预测精度高,MSE平均值分别低0.01和0.26,MAE平均值分别低0.05和0.12。结果表明,所提出的方法能够有效监控飞控系统,为无人机飞行管理决策提供数据支持。
Monitoring of condition parameters of flight control system is an important means to ensure the flight safety of UAVs.According to the composition characteristics and flight control law of the UAV flight control system,a flight control system condition monitoring model based on LSTM is designed and constructed.Firstly,the UAV historical flight parameter data training model is used to establish the regression mapping relationship between condition parameters and input flight data.Then,the trained network model is used to predict the state parameters of flight control system in real time.By comparing the difference between the measured value and the predicted value,the condition monitoring of the flight control system is realized.The flight data of UAV are selected for experiments,and the algorithm based on LSTM has higher prediction accuracy than BPNN and SVM,with average values of MSE lower than 0.01 and 0.26,and average values of MAE lower than 0.05 and 0.12,respectively.The results show that the proposed method can effectively monitor the flight control system and provide data support for UAV flight management decision-making.
作者
王凤芹
高龙
李瑛
耿宝亮
WANG Fengqin;GAO Long;Li Ying;GENG Baoliang(Naval Aviation University,Yantai Shandong 264001,China)
出处
《海军航空大学学报》
2022年第5期387-392,共6页
Journal of Naval Aviation University
基金
国家自然科学基金(51605487)。
关键词
飞控系统状态监控
无人机飞行安全
状态监控算法
深度学习
飞参数据分析
flight control system condition monitoring
UAV flight safety
condition monitoring algorithm
deep learning
flight parameter data analysis
