航空发动机健康估计的神经网络修正卡尔曼滤波算法

推进技术 ›› 2018, Vol. 39 ›› Issue (11): 2564-2570.

航空发动机健康估计的神经网络修正卡尔曼滤波算法

顾嘉辉,黄金泉,鲁峰

南京航空航天大学能源与动力学院，江苏省航空动力系统重点实验室，江苏南京 210016,南京航空航天大学能源与动力学院，江苏省航空动力系统重点实验室，江苏南京 210016,南京航空航天大学能源与动力学院，江苏省航空动力系统重点实验室，江苏南京 210016

发布日期:2021-08-15
作者简介:顾嘉辉，男，硕士生，研究领域为航空发动机建模与故障诊断。E-mail:gujiahui419@126.com 通讯作者:黄金泉，男，博士，教授，研究领域为航空发动机建模、控制与故障诊断。

Neural Network Corrected Kalman Filter Algorithm for Aero-Engine Health Parameters Estimation

Jiangsu Province Key Laboratory of Aerospace Power System，College of Energy and Power Engineering，Nanjing University of Aeronautics and Astronautics，Nanjing 210016，China,Jiangsu Province Key Laboratory of Aerospace Power System，College of Energy and Power Engineering，Nanjing University of Aeronautics and Astronautics，Nanjing 210016，China and Jiangsu Province Key Laboratory of Aerospace Power System，College of Energy and Power Engineering，Nanjing University of Aeronautics and Astronautics，Nanjing 210016，China

Published:2021-08-15

摘要/Abstract

摘要： 针对商用航空发动机与气路相关的传感器分布不均、且个数小于气路健康参数的个数、使用卡尔曼滤波算法估计全部气路健康参数时容易出现误判的特点，提出一种神经网络修正的卡尔曼滤波算法。该算法在每个采样周期内利用BP神经网络来修正个体的偏移方向，按粒子滤波算法计算每个个体的权值用以估计总体的均值和协方差，然后利用卡尔曼滤波算法更新所有个体，并将总体的均值作为当前时刻的估计结果。通过对商用航空发动机部件级模型在多个飞行状态点数字仿真模拟9种气路突变故障，由7个可测输出估计全部10个健康参数，该混合算法的估计误差相比BP神经网络与无迹卡尔曼滤波算法分别平均降低了34.6%与47.9%。

关键词: 商用航空发动机；故障诊断；神经网络；卡尔曼滤波；粒子滤波

Abstract: Aiming at the fact that Kalman filter is apt to misjudge the gas path health parameters of a commercial aircraft engine since available on-board sensors are unevenly distributed and the number of them is usually less than the number of health parameters， a Neural Network corrected Kalman Filter (NN-KF) algorithm is proposed. Individuals of the algorithm are corrected by Back Propagation-Neural Network (BP-NN) and their weights are calculated on the basis of Particle Filter (PF) during each sampling period in order to estimate mean and covariance， then KF is adopted to update individuals. The mean vector is regarded as estimated result every time step. Several types of abrupt gas path faults of a commercial aircraft engine at multiple fly points have been numerically simulated and all 10 health parameters are estimated based on 7 measured outputs. The simulation results suggest that the hybrid method compared with BP-NN and Unscented Kalman Filter (UKF) reduces estimated errors by averages of 34.6% and 47.9%， respectively.

Key words: Commercial aircraft engine；Fault diagnosis；Neural network；Kalman filter；Particle filter

顾嘉辉,黄金泉,鲁峰. 航空发动机健康估计的神经网络修正卡尔曼滤波算法[J]. 推进技术, 2018, 39(11): 2564-2570.

GU Jia-hui,HUANG Jin-quan,LU Feng. Neural Network Corrected Kalman Filter Algorithm for Aero-Engine Health Parameters Estimation[J]. Journal of Propulsion Technology, 2018, 39(11): 2564-2570.

参考文献

[1] Mercer C R, Simon D L, Hunter G W, et al. Fundamental Technology Development for Gas-Turbine Engine Health Management[R]. NASA/TM-2007-0022364.
[2] Litt J S, Simon D L, Gary S, et al, A Survey of Intelligent Control and Health Management Technologies for Aircraft Propulsion Systems[R]. NASA/TM-2005-213622.
[3] Kobayashi T, Simon D L. Integration of On-Line and Off-Line Diagnostic Algorithms for Aircraft Engine Health Management[J]. Journal of Engineering for Gas Turbines & Power, 2007, 129(4): 649-659.
[4] 薛薇, 郭迎清, 李睿. 航空发动机状态监视、故障诊断研究及验证[J]. 推进技术, 2011, 32(2): 271-275. (XUE Wei, GUO Ying-qing, LI Rui. Algorithm and Experimental Validation for Condition Monitoring, Fault Detection for Gas Turbine Engine[J]. Journal of Propulsion Technology, 2011, 32(2): 271-275.)
[5] Zhou G T, Su S M. Application of Improved Extended Kalman Filter for Adaptive Model of Aircraft Engine[J]. Applied Mechanics & Materials, 2015, 709: 180-185.
[6] 陆军, 郭迎清, 张书刚. 基于改进混合卡尔曼滤波器的航空发动机机载自适应模型[J]. 航空动力学报, 2011, 26(11): 2593-2600.
[7] 陈煜, 鞠红飞, 鲁峰, 等. 涡喷发动机健康状态的带约束非线性滤波估计[J]. 推进技术, 2016, 37(4): 741-748. (CHEN Yu, JU Hong-fei, LU Feng, et al. Nonlinear Filter with Inequality Constraints for Turbojet Engine Health Estimation[J]. Journal of Propulsion Technology, 2016, 37(4): 741-748.)
[8] 姜洁, 李秋红, 张高钱, 等. 基于NN-ELM的航空发动机燃油系统执行机构故障诊断[J]. 航空动力学报, 2016, 31(2): 484-492.
[9] 尤成新, 鲁峰, 黄金泉. 航空发动机传感器信号重构的K-ELM方法[J]. 航空动力学报, 2017, 32(1):221-226.
[10] 李长征, 张瑜. 基于广义回归神经网络的传感器故障检测[J]. 推进技术, 2017, 38(9): 2130-2137. (LI Chang-zheng, ZHANG Yu. Sensor Fault Detection Based on General Regression Neural Network[J]. Journal of Propulsion Technology, 2017, 38(9): 2130-2137.)
[11] 江传尚, 樊丁, 马冲. 小波网络在某型航空发动机故障诊断中的应用[J]. 航空动力学报, 2009, 24(4): 892-895.
[12] Julier S J, Uhlmann J K. Unscented Filtering and Nonlinear Estimation[J]. Proceedings of the IEEE, 2004, 92(3): 401-422.
[13] 朱志宇. 粒子滤波算法及其应用[M]. 北京:科学出版社, 2010.
[14] Hecht-Nielsen. Theory of the Backpropagation Neural Network[C]. Washington: International Joint Conference on Neural Networks, 1989.
[15] Demuth H, Beale M. Neural Network Toolbox-For Use with MATLAB[J]. Matlab Users Guide the Math Works, 1995, 21(15): 1225-1233.
[16] Kobayashi T, Simon D L. Hybrid Kalman Filter: A New Approach for Aircraft Engine In-Flight Diagnostics[R]. NASA/TM-2006-214491.
[17] Moré J J. The Levenberg-Marquardt Algorithm: Implementation and Theory[M]. Berlin: Springer, 1978.
[18] Simon D, Simon D L. Aircraft Turbofan Engine Health Estimation Using Constrained Kalman Filtering[J]. Journal of Engineering for Gas Turbines & Power, 2003, 127(2).
[19] 胡卫红, 李述清, 孙健国. 控制问题中航空发动机飞行包线区域最优划分[J]. 推进技术, 2011, 32(3):391-395. (HU Wei-hong, LI Shu-qin, SUN Jian-guo.Flight-Envelope Optimization Partition for Aeroengines Control[J]. Journal of Propulsion Technology, 2011, 32(3): 391-395.)　　　　　　　　　　　　　　收稿日期:2017-10-13；修订日期:2017-12-27。作者简介:顾嘉辉,男,硕士生,研究领域为航空发动机建模与故障诊断。E-mail:gujiahui419@126.com通讯作者:黄金泉,男,博士,教授,研究领域为航空发动机建模、控制与故障诊断。E-mail: jhuang@nuaa.edu.cn(编辑:史亚红)