航空发动机智能控制与健康管理专栏

基于非线性关联挖掘的航空发动机气路故障诊断方法

  • 詹轲倚 ,
  • 廖增步 ,
  • 冷文龙 ,
  • 潘一 ,
  • 宋志平 ,
  • 黄维娜
展开
  • 1.中国航发贵阳发动机设计研究所,贵阳 550081
    2.贵州省转子结构完整性重点实验室,贵阳 550081
    3.清华大学 航空发动机研究院,北京 100089
    4.浙江大学 航空航天学院,杭州 310007
    5.航空动力系统与等离子体技术全国重点实验室,西安 710049
    6.西安交通大学 机械工程学院,西安 710049
.E-mail: Zengbu.Liao@zju.edu.cn

收稿日期: 2026-01-05

  修回日期: 2026-03-06

  录用日期: 2026-04-15

  网络出版日期: 2026-04-30

基金资助

贵州省科技计划项目;中国航发集团自主创新基金(ZZCX-2024-0062)

Aero-engine gas path fault diagnosis method based on nonlinear correlation mining

  • Keyi ZHAN ,
  • Zengbu LIAO ,
  • Wenlong LENG ,
  • Yi PAN ,
  • Zhiping SONG ,
  • Weina HUANG
Expand
  • 1.AECC Guiyang Engine Research Institute,Guiyang 550081,China
    2.Guizhou Province Key Laboratory of Rotor Structural Integrity,Guiyang 550081,China
    3.Institute for Aero Engine,Tsinghua University,Beijing 100089,China
    4.School of Aeronautics and Astronautics,Zhejiang University,Hangzhou 310007,China
    5.National Key Lab of Aerospace Power System and Plasma Technology,Xi’an 710049,China
    6.School of Mechanical Engineering,Xi’an Jiaotong University,Xi’an 710049,China

Received date: 2026-01-05

  Revised date: 2026-03-06

  Accepted date: 2026-04-15

  Online published: 2026-04-30

Supported by

AECC Independent Innovation Fund(ZZCX-2024-0062);Guizhou Science and Technology Program

摘要

气路故障诊断是航空发动机健康管理系统的重要组成部分,其关键在于故障特征提取。近年来,随着深度学习技术的发展,基于图神经网络的气路故障特征提取方法受到广泛关注。然而,传统图神经网络只能提取节点之间的加权求和关系,而忽略了发动机气路系统中普遍存在的非线性耦合关系,导致跨工况诊断准确性和可解释性不足。为此,提出了一种基于非线性关联挖掘的气路故障诊断方法。该方法通过原创的非线性关联挖掘层和改进的图卷积层实现了气路故障特征的可解释提取。通过500架次全寿命期仿真数据对其性能进行了测试验证,所提方法全程未发生虚警,检测率为88.99%,隔离率为98.61%,在收敛性和诊断准确性等方面均显著优于卷积、图卷积、图注意力网络等对比方法。

本文引用格式

詹轲倚 , 廖增步 , 冷文龙 , 潘一 , 宋志平 , 黄维娜 . 基于非线性关联挖掘的航空发动机气路故障诊断方法[J]. 航空学报, 2026 , 47(15) : 633321 -633321 . DOI: 10.7527/S1000-6893.2026.33321

Abstract

Gas path fault diagnosis is an essential component of aero-engine health management systems, with fault feature extraction being its key aspect. In recent years, with the advancement of deep learning techniques, gas path fault feature extraction methods based on graph neural networks have attracted considerable attention. However, conventional graph neural networks can only capture linear weighted aggregation relationships among nodes while neglecting the nonlinear coupling relationships prevalent in engine gas path systems, resulting in insufficient cross-condition diagnostic accuracy and interpretability. To address this issue, a gas path fault diagnosis method based on nonlinear correlation mining is proposed. This method achieves interpretable extraction of gas path fault features through an original nonlinear correlation mining layer and an improved graph convolutional layer. The performance of the proposed method was validated using full-lifecycle simulation data encompassing 500 flight sorties. The proposed method achieved zero false alarms throughout the entire operational period, with a detection rate of 88.99% and an isolation rate of 98.61%, significantly outperforming comparative methods based on convolutional, graph convolutional, and graph attention networks in terms of convergence and diagnostic accuracy.

参考文献

[1] 黄维娜, 黎方娟, 祁宏斌. 航空发动机数字工程初步研究与发展思考[J]. 航空学报202445(5): 529693.
  HUANG W N, LI F J, QI H B. Preliminary investigation and thoughts on aero-engine digital engineering development[J]. Acta Aeronautica et Astronautica Sinica202445(5): 529693 (in Chinese).
[2] MEHER-HOMJI C B, CHAKER M, BROMLEY A F. The fouling of axial flow compressors: Causes, effects, susceptibility, and sensitivity[C]∥ASME Turbo Expo 2009: Power for Land, Sea, and Air. New York: ASME, 2009: 571-590.
[3] GANNAN A. Cascade testing and CFD applied to gas turbine performance improvement with compressor cleaning[D]. Bedford: Cranfield University, 2010: 57.
[4] METWALLY M, TABAKOFF W, HAMED A. Blade erosion in automotive gas turbine engine[J]. Journal of Engineering for Gas Turbines and Power1995117(1): 213-219.
[5] EJAZ N, QURESHI IN, RIZVI SA. Creep failure of low pressure turbine blade of an aircraft engine[J]. Engineering Failure Analysis201118(6):1407-1414.
[6] IATA. Maintenance costs for aging aircraft[M]. Montreal: International Air Transport Association, 2018.
[7] LAPESA BARRERA D. AMP content and maintenance planning document (MPD)[M]∥Aircraft Maintenance Programs. Cham: Springer International Publishing, 2022:33-46.
[8] 廖增步, 张瑞, 耿佳, 等. 小涵道比涡扇发动机机载气路故障诊断研究综述: 边界条件与研究难点[J]. 推进技术202445(12): 24-40.
  LIAO Z B, ZHANG R, GENG J, et al. A review on onboard gas-path fault diagnosis for low bypass ratio turbofan engines: Boundaries and challenges[J]. Journal of Propulsion Technology202445(12): 24-40 (in Chinese).
[9] URBAN L A. Gas path analysis applied to turbine engine condition monitoring[J]. Journal of Aircraft197310(7): 400-406.
[10] LUPPOLD R, ROMAN J, GALLOPS G, et al. Estimating in-flight engine performance variations using Kalman filter concepts[C]∥ 25th Joint Propulsion Conference. Reston: AIAA, 1989.
[11] 李国庆. 人工智能方法在航空发动机故障诊断中的应用研究[D]. 沈阳: 沈阳航空航天大学, 2019: 2.
  LI G Q. Application research of artificial intelligence methods in aero-engine fault diagnosis[D]. Shenyang: Shenyang Aerospace University, 2019: 2 (in Chinese).
[12] 吴超, 陈磊, 刘渊, 等. 基于特征优化和支持向量机的航空发动机气路故障诊断[J]. 航空发动机202450(4): 30-37.
  WU C, CHEN L, LIU Y, et al. Aeroengine gas-path fault diagnosis based on feature optimization and support vector machine[J]. Aeroengine202450(4): 30-37 (in Chinese).
[13] ZHONG S S, FU S, LIN L. A novel gas turbine fault diagnosis method based on transfer learning with CNN[J]. Measurement2019137: 435-453.
[14] XU C Y, GUI X C, ZHAO Y. Digital twin-assisted multiview reconstruction enhanced domain adaptation graph networks for aero-engine gas path fault diagnosis. IEEE Sensors Journal202424(13):21694-21705.
[15] 曹力昂, 于博升, 张会生. 基于层次自适应特征提取的航空发动机气路故障诊断[J/OL]. 上海交通大学学报, (2025-04-29)[2025-12-20]. .
  CAO L A, YU B S, ZHANG H S. Hierarchical adaptive feature extraction-based gas path fault diagnosis for aero-engine[J/OL]. Journal of Shanghai Jiao Tong University, (2025-04-29)[2025-12-20]. (in Chinese).
[16] 张世杰, 胡家文, 苗国磊. 基于工况识别与自训练时空图卷积的航空发动机气路故障诊断[J]. 推进技术202445(11): 251-260.
  ZHANG S J, HU J W, MIAO G L. Fault diagnosis of aero-engine gas path based on condition recognition and self-training ST-GCN model[J]. Journal of Propulsion Technology202445(11): 251-260 (in Chinese).
[17] BAO P, YI W, ZHU Y, et al. STHFD: Spatial-temporal hypergraph-based model for aero-engine bearing fault diagnosis[J]. Aerospace202512(7): 612.
[18] SHAH B, SARVAJITH M, SANKAR B, et al. Multi-auto associative neural network based sensor validation and estimation for aero-engine[C]∥2013 IEEE AUTOTESTCON. Piscataway: IEEE, 2013: 1-7.
[19] LI T, ZHAO Z, SUN C, et al. Hierarchical attention graph convolutional network to fuse multi-sensor signals for remaining useful life prediction[J]. Reliability Engineering & System Safety2021215: 107878.
[20] XU C Y, GUI X C, HUO M X. Domain adversarial enhanced multi-channel graph networks for aeroengine gas path fault diagnosis[C]∥IECON 2023-49th Annual Conference of the IEEE Industrial Electronics Society. Piscataway: IEEE, 2023: 1-6.
[21] GOYAL V. Anomaly detection and failure prediction in gas turbines[D]. Orlando: University of Central Florida, 2021: 14.
[22] HUANG Y F, TAO J, ZHAO J Y, et al. Graph structure embedded with physical constraints-based information fusion network for interpretable fault diagnosis of aero-engine[J]. Energy2023283:129120.
[23] ZHANG S M, SONG P Y, ZHAO C H. Multi-conditional adjacency learning and grouping method for monitoring gas turbine generation process with nonstationarity[C]∥2023 China Automation Congress (CAC). Piscataway: IEEE, 2024.
[24] CHENG K R, ZHANG K Y, WANG Y Z, et al. Research on gas turbine health assessment method based on physical prior knowledge and spatial-temporal graph neural network[J]. Applied Energy2024367: 123419.
[25] CHEN Y, LIANG C, LIU D, et al. Embedding-graph neural-network for transient NO x emissions prediction[J]. Energies202316(1): 3.
[26] WANG Z, FU X, ZHANG R, et al. Knowledge and data jointly driven aeroengine gas path performance assessment method[J]. Chinese Journal of Aeronautics202437(5): 533-557.
[27] LU F, LI Z, HUANG J, et al. Hybrid state estimation for aircraft engine anomaly detection and fault accommodation[J]. AIAA Journal202058 (4): 1748-1762.
[28] YANG K, TU Q, ZENG Y, et al. Parameter selection for aeroengine transient state gas path analysis[J]. Journal of the Global Power and Propulsion Society20226: 61-73.
[29] VELI?KOVI? P, CUCURULL G, CASANOVA A, et al. Graph attention networks[C]∥6th International Conference on Learning Representations. 2018.
[30] 匡育衡, 王正宁, 王正, 等. 基于双注意力时空图卷 积神经网络的4D轨迹预测方法[J]. 电子科技大学学报202554(5): 641-651.
  KUANG Y H, WANG Z N, WANG Z, et al. 4D trajectory prediction based on dual-attention spatiotemporal graph convolutional neural network[J]. Journal of University of Electronic Science and Technology of China202554(5): 641-651 (in Chinese).
[31] JIANG K, ZENG H, WU Z, et al. Study on the effect of parameter sensitivity on engine optimization results[J]. Energies202316(23): 7899.
[32] LIAO Z B, WANG J, LIU J X, et al. Uncertainties in gas path diagnosis of gas turbines: Representation and impact analysis[J]. Aerospace Science and Technology2021113: 106724.
[33] SALLEE G P. Performance deterioration based on existing (historical) data; JT9D jet engine diagnostics program: NASA-CR-135448[R]. Washington, D.C.: NASA, 1978.
[34] LIAO Z B, ZHAN K Y, ZHAO H, et al. Addressing class imbalanced learning in real-time aero-engine gas-path fault diagnosis via feature filtering and mapping[J].Reliability Engineering & System Safety2024249: 110189.
文章导航

/