摘要
针对飞行训练安全,大规模训练数据更新时预测准确性较差等问题,研究提出一种基于增量学习的在线飞行训练姿态预测模型。模型针对飞行训练数据进行相关性分析,使用基于流数据的增量学习模型对飞行训练数据中的俯仰角进行预测研究。实验结果表明,模型在对飞行训练俯仰角进行预测时,预测准确率高,误差小。通过模型对比实验,模型相较于传统预测模型,精度高、误差小、泛化能力强,可以有效预测飞行数据,保障飞行安全。
Aiming at the safety of flight training and the poor prediction accuracy when large scale training data is updated,an online flight training attitude prediction model based on incremental learning is proposed.The model performs correlation analysis on flight training data,and uses the incremental learning model based on streaming data to predict the pitch angle in flight training data.The experimental results show that the model has high prediction accuracy and small error when predicting the pitch angle of flight training.Through model comparison experiments,compared with traditional prediction models,the model has high accuracy,small error,and strong generalization ability,which can effectively predict flight data and ensure flight safety.
作者
路晶
史宇
任洲
LU Jing;SHI Yu;REN Zhou(Civil Aviation Flight University of China,Guanghan 618000,China;Nanjing University of Aeronautics and Astronautics,Nanjing 211000,China)
出处
《航空计算技术》
2022年第5期5-8,共4页
Aeronautical Computing Technique
基金
四川省重点研发项目资助(2022YFG0027)
民航飞行技术与飞行安全重点实验室自主研究项目资助(FZ2020ZZ02)
民航局安全能力项目资助(14002600100020J103)。
关键词
训练安全
飞行姿态
时序预测
增量学习
training safety
flight attitude
time series prediction
incremental learning