摘要
针对航空发动机剩余寿命预测中深度学习算法参数优化效率低、预测准确率差等问题,在长短期记忆网络算法(LSTM)的基础上提出一种基于贝叶斯优化的LSTM算法。利用长短期记忆网络对航空发动机的传感器数据进行时间序列预测,并运用贝叶斯优化算法对长短期记忆网络的超参数进行迭代优化。利用NASA公开数据集对算法进行验证,结果表明,相较于其他算法,优化后的算法在优化参数、提高预测准确率方面有明显改善,能为保证航空器的安全使用以及制定维修替换策略提供参考。
Aiming at the problems of low efficiency of parameter optimization and poor prediction accuracy of deep learning algorithm in aero-engine residual service life prediction,based on long-term and short-term memory network algorithm(LSTM),a new LSTM algorithm based on Bayesian optimization is proposed.The long-term and short-term memory network is used to predict the time series of aeroengine sensor data,and the Bayesian optimization algorithm is used to iteratively optimize the super parameters of the long-term and short-term memory network.The algorithm is verified with NASA public data sets.The results show that the optimized algorithm has obvious improvements in optimizing parameters and improving prediction accuracy compared with other algorithms,which can provide reference for ensuring the safe use of aircraft and formulating maintenance replacement strategy.
作者
张其霄
董鹏
王科文
卢苇
ZHANG Qi-xiao;DONG Peng;WANG Ke-wen;LU Wei(Department of Management Engineering and Equipment Economics,Naval University of Engineering,Wuhan 430033,China)
出处
《火力与指挥控制》
CSCD
北大核心
2022年第4期85-89,共5页
Fire Control & Command Control
