摘要
为建立一种适用于大包线、变状态的高精度、高实时性航空发动机机载自适应稳态模型,提出一种基于神经网络和推进系统矩阵相融合(NN-PSM)的机载自适应稳态模型建模方法。该方法基于小偏差线性化方法对发动机进行线性化来提取推进系统矩阵,用于表征机载模型与发动机之间的输出偏差量。基于神经网络建立发动机基线模型,用于映射飞行条件与发动机输出量之间的关系,利用神经网络的强拟合能力提高机载模型的稳态精度;设计卡尔曼滤波器实时估计发动机健康参数,提高模型的自适应能力。在大包线、变状态的飞行条件下进行仿真验证,并与传统的复合推进系统模型(CPSM)进行对比,结果表明:NN-PSM模型的平均精度在0.66%以内,而CPSM的平均精度为2.07%以内,运行时间仅为CPSM的1/10,且具有数据存储量少的特点。
In order to establish a high-precision,high-real-time aero-engine on-board adaptive steady-state model suitable for large envelopes and multiple states, a on-board adaptive steady-state model based on neural network and propulsion system matrix fusion(NN-PSM)was proposed. Adaptive steady-state modeling method was based on small deviation linearization method to linearize the engine to extract the propulsion system matrix,which was used to characterize output deviation. The engine baseline model was established based on the neural network,and the relationship between the flight conditions and the engine output was mapped, and the neural network used the strong fitting ability to improve the steady-state accuracy of the on-board model. The Kalman filter was designed in real time to improve the adaptive ability of the model.The simulation was carried out under large envelope and variable state flight conditions, and compared with the traditional compact propulsion system model(CPSM) model. The results showed that the average accuracy of the NN-PSM model was within 0. 66%,while the average accuracy of the CPSM model was about 2. 07%;the time was about one-tenth of the CPSM model,and the amount of data storage was small.
作者
项德威
郑前钢
张海波
陈铖
房娟
XIANG Dewei;ZHENG Qiangang;ZHANG Haibo;CHEN Cheng;FANG Juan(Jiangsu Province Key Laboratory of Aerospace Power System,College of Energy and Power Engineering,Nanjing University of Aeronautics and Astronautics,Nanjing 210016,China)
出处
《航空动力学报》
EI
CAS
CSCD
北大核心
2022年第2期409-423,共15页
Journal of Aerospace Power
基金
国家自然科学基金(51906102)
国家科技重大专项(2017-V-0004-0054)
智能航空发动机基础问题研究项目(2017-JCJQ-ZD-047-21)
中央高校基本科研业务费(NZ2020002)。
关键词
机载稳态模型
推进系统矩阵
神经网络
基线模型
自适应
on-board steady-state model
propulsion system matrix
neural network
baseline-model
adaptive
