基于LSTMPI-GNN预测燃烧室涡轮耦合组件多物理场的方法

李宇轩; 刘伟杭; 刘海涛; 崔苗; 任佳文

doi:10.7527/S1000-6893.2026.33578

航空学报 >

0 1 - 0

DOI: https://doi.org/10.7527/S1000-6893.2026.33578

基于LSTMPI-GNN预测燃烧室涡轮耦合组件多物理场的方法

李宇轩 ,
刘伟杭 ,
刘海涛 ,
崔苗 ,
任佳文

展开

大连理工大学

收稿日期: 2026-03-17

修回日期: 2026-05-22

网络出版日期: 2026-05-28

基金资助

国家自然科学基金;中央高校基本科研业务费理科基础项目

收起

A approach for estimating multi-physics fields of combustor-turbine cross-components based on LSTMPI-GNN

LI Yu-Xuan ,
LIU Wei-Hang ,
LIU Hai-Tao ,
CUI Miao ,
REN Jia-Wen

Expand

Received date: 2026-03-17

Revised date: 2026-05-22

Online published: 2026-05-28

Fold

摘要

针对航空发动机燃烧室涡轮耦合组件复杂瞬态燃烧过程的高效、高精度预测需求，本文提出一种融合图神经网络 (GNN)、长短期记忆网络 (LSTM)与物理信息神经网络 (PINN)的LSTMPI?GNN混合建模框架。该框架旨在协同发挥各组件的优势：首先利用GNN从非结构网格的燃烧场数据中有效捕获复杂的空间拓扑与局部特征；进而通过LSTM对燃烧室状态的动态演化过程进行时序建模，以记忆并学习其时间依赖特征；最后，引入PINN将描述燃烧与流动过程的大涡模拟控制方程作为物理约束，对神经网络预测结果进行正则化与修正，从而增强模型的物理一致性与泛化能力。该框架基于有限时段的大涡模拟数据进行训练，最终实现了对燃烧室涡轮耦合组件在连续时间步下三维多物理场的快速预测。结果表明，所建模型对温度场、速度场及燃烧产物分布的最大预测误差低于7%，并展现出约两倍于训练时长的有效外推能力。在完成一次性训练后，模型对新时刻的预测可在秒级内完成，显著提升了燃烧室涡轮耦合组件多工况分析与优化设计的效率。本研究为燃烧场的高效预测与动态预报提供了一种兼具高精度、快速响应与良好泛化能力的解决方案。

关键词： 航空发动机; 燃烧室; 大涡模拟; 物理信息神经网络; 多物理场

本文引用格式

李宇轩 , 刘伟杭 , 刘海涛 , 崔苗 , 任佳文 . 基于LSTMPI-GNN预测燃烧室涡轮耦合组件多物理场的方法[J]. 航空学报, 0 : 1 -0 . DOI: 10.7527/S1000-6893.2026.33578

Abstract

To address the need for efficient and high-precision prediction of complex transient combustion processes in aeroengine combustor-turbine cross-components, this paper proposes a hybrid LSTMPI-GNN modeling framework that integrates Graph Neural Networks (GNN), Long Short-Term Memory networks (LSTM), and Physics-Informed Neural Networks (PINN). The proposed framework is designed to synergistically leverage the strengths of each component: the GNN effectively captures complex spatial topologies and local features from unstructured combustion field data; the LSTM subsequently models the temporal evolution of the combustor state, learning and memorizing its time-dependent characteristics; finally, the PINN incorporate the governing equations of Large Eddy Simulation (LES), which describe the combustion and flow processes, as physical constraints to regularize and correct the neural networks’ predictions, thereby enhancing the model's physical consistency and generalization capability. By training limited-duration LES data, this framework achieves rapid prediction of three-dimensional multi-physics fields for the combustor-turbine component across consecutive time steps. The results demonstrate that the model maintains maximum prediction errors below 7% for temperature, velocity, and combustion product fields. Furthermore, it exhibits effective extrapolation capability for a duration approximately twice the length of its training sequence. After a one-time training, the model can generate predictions for new time instances within seconds, significantly improving the efficiency of multi-condition analysis and optimized design for combustor-turbine components. This study provides a solution for efficient prediction of combustion fields and dynamic forecasting, offering high accuracy, rapid response, and robust generalization.

Key words： Aeroengine; Combustor; Large eddy simulation; Physics-informed neural network; Multi-physics field

参考文献

[1] WANG X, KONG C, REN M H, et al. Research on temperature field prediction method in an aero-engine combustor with high generalization ability, Applied Thermal Engineering [J], 2024, 239: 122042.
[2] 张伟伟，寇家庆，刘溢浪. 智能赋能流体力学展望[J]. 航空学报，2021，42(4): 524689. ZHANG W W, KOU J Q, LIU Y L. Prospect of artifi-cial intelli-gence empowered fluid mechanics [J]. Acta Aero-nauticaet Astronautica Sinica, 2021, 42(4): 524689. (in Chinese)
[3] KRULJEVIC B, WANG Q S, VAYSSE N, et al. Nu-merical study on the impact of Nanosecond Repeti-tively Pulsed discharges on the lean blowout limit for a hydrogen/air swirled flame, Combustion and Flame [J], 2025, 277: 114149.
[4] LIU X C, ZHAO W L, GUO H Q, et al. XCT damage evaluation and analysis of SiC turbine guide vanes af-ter thermal shocks, International Journal of Applied Ceramic Technology [J], 2023, 20: 2919.
[5] CHEN Y Q, FAN Y X, XU L L, et al. Experimental and numerical study of flow and ignition and lean blowout characteristics of jet-cooled wall flameholder in a dual-mode combustor, Aerospace Science and Technology [J], 2022, 122: 107403.
[6] HU G, QIN Q Y, JIN W, et al. Large Eddy Simulation of the Influences of the Pilot-Stage Structure on the Flow Characteristics in a Centrally Staged High-Temperature-Rise Combustor, Journal of Aerospace Engineering [J], 2022, 9: 782.
[7] PATHANIA R S, SKIBA W A, EPAMINONDAS M, et al. Blow-off mechanisms of turbulent premixed bluff-body stabilised flames operated with va-pourised kerosene fuels, Proceedings of the Combus-tion Institute [J], 2023, 140: 108413.
[8] SUN B, WANG Y Y, SHI C, et al. Combined effect of chamber geometry and compression ratio on the combustion process in a methanol-enriched rotary engine, Energy [J], 2025, 318: 134772.
[9] BAN Y Y, ZHANG F, ZHANG Z H, et al. A compar-ative numerical study of plasma and spark assisted ignition in a cavity-based supersonic combustor, Combustion and Flame [J], 2025, 318: 134772.
[10] 向安定, 刘松龄, 朱惠人. 涡轮叶片型面气膜冷却效率的计算模型[J]. 推进技术, 2004, 04: 303-306.
XIANG A D, LIU S L, ZU H R, et al. A new model for film cooling effectiveness computation on blade surface in a turbine cascade[J]. Journal of Propulsion Technology, 2004, 15(4): 303-306. (in Chinese)
[11] MATAR C, CINNELLA P, GLOERFELT X, et al. Cost-effective multi-fidelity strategy for the optimiza-tion of high-Reynolds number turbine flows guided by LES, Aerospace Science and Technology [J], 2025, 164: 110426.
[12] CHEN Z S, WU Y J, XU Y M, et al. Field-validated LES study on the aerodynamic response of a long-span bridge in complex double-mountain terrain, En-gineering Structures [J], 2025, 343: 2025.
[13] 程穆威, 杨晓光, 范永升, 等. 基于LSTM的航空发动机涡轮盘局部服役温度-应力快速预测方法[J]. 航空动力学报, 2025, 40(10)：20240547.
CHENG M W, YANG X G, FAN Y S, et al. Rapid prediction method for local service temperature and stress on aeroengine turbine disks based on LSTM network[J]. Journal of Aerospace Power, 2025, 40(10)：20240547 (in Chinese).
[14] 邢素霞, 李珂娴, 方俊泽, 等. 深度学习下的医学图像分割综述[J]. 计算机工程与应用, 2025, 61(07): 25-41.
XING S X, LI K X, FANG J Z, et al. Survey of Medical Image Segmentation in Deep Learning [J]. Computer Engineering and Applications, 2025, 61 (7): 25-41. (in Chinese).
[15] ZHANG S H, DING S T, LIU P , et al. Effects of swirl and hot streak on thermal performances of a high-pressure turbine, Chinese Journal of Aeronautics [J], 2023, 36: 250-267.
[16] SAINI P, CHRISTOPH M, STEINBERG A M. Devel-opment and evaluation of gappy-POD as a data re-construction technique for noisy PIV measurements in gas turbine combustors, Experiments in Fluids[J], 2016, 57(122): 1–15.
[17] LI X, NIOU X T, LIU Z Y, et al. A hybrid graph neu-ral network-based framework for aerothermal predic-tion on hypersonic vehicles, Physics of Fluids [J], 2025, 37: 077163.
[18] XU M Q, YE T, YANG M T, et al. Combustion flow field reconstruction in a hydrogen-fueled scramjet combustor based on lightweight generative adversarial networks, International Communications in Heat and Mass Transfer [J], 2025, 165: 109036.
[19] HUANG S C, YE T, DENG X. Supersonic combus-tion flow field reconstruction in a scramjet based on deep learning method, Aerospace Science and Tech-nology [J], 2025, 165: 109036.
[20] GUO M M, DENG X, MA Y, et al. Hypersonic inlet flow field reconstruction dominated by shock wave and boundary layer based on small sample physics-informed neural networks, Aerospace Science and Technology [J], 2024, 150: 109205.
[21] SUN L H, YANG B Q, MA J, et al. Trajectory predic-tion in pipeline form for intercepting hypersonic glid-ing vehicles based on LSTM, Chinese Journal of Aeronautics [J], 2023, 36: 421-433.
[22] LI G Z, ZHANG S, ZHOU S, et al. Large eddy simu-lations of the turbine vane pressure side film cooling flows of cylindrical and fan-shaped holes with a saw-tooth plasma actuator[J]. Applied Thermal Engineer-ing, 2024, 257(1): 124404.
[23] 黄渭清, 李宁, 刘开霖, 等. 基于MAML-LSTM的服役涡轮叶片损伤状态评估方法[J].航空学报, 2024, 45(18): 275-286.
HUANG W Q, LI N, LIU K, et al. Damage assessment method for in-service turbine blades based on MAML-LSTM [J]. Acta Aeronauticaet Astronautica Sinica, 2024, 45 (18): 275-286(in Chinese).
[24] 吴跃腾, 巴顿, 杜娟, 等. 基于深度注意力网络的压气机流场重构方法[J].航空学报,2024, 45(24): 85-99.
WU Y T, BA D, DU J, et al. Reconstruction method of compressor flow field based on deep attention network [J]. Acta Aeronauticaet Astronautica Sinica, 2024, 45 (24): 85-99(in Chinese).
[25] SOY S, SARKAR D R, ROY P, et al. Adaptive Inter-face-PINNs (AdaI-PINNs) for transient diffusion: Applications to forward and inverse problems in het-erogeneous media, Finite Elements in Analysis & De-sign[J], 2025, 244: 104305.
[26] ZHANG R F, LI X C, WANG N, et al. Enhancing complex Fourier characterization for temperature field reconstruction via multi-scale modulation and demodulation[J]. International Journal of Thermal Sciences, 2025 211: 109694.

Options

文章导航

模态框（Modal）标题

摘要

本文引用格式

Abstract

参考文献