基于数据驱动的可控变形叶型优化方法

龙嘉明; 潘天宇; 李宸璋; 郑孟宗; 李秋实

doi:10.13224/j.cnki.jasp.20220779

基于数据驱动的可控变形叶型优化方法

doi: 10.13224/j.cnki.jasp.20220779

龙嘉明^1,,
潘天宇^{1, 2,*, ,},
李宸璋³,
郑孟宗¹,
李秋实^{1, 4}

1.
北京航空航天大学航空发动机研究院，北京 100191
2.
中国航空发动机研究院先进航空动力创新工作站，北京 101300
3.
北京航空航天大学能源与动力工程学院，北京 100191
4.
西华大学流体与动力机械重点实验室，成都 610039

基金项目: 国家自然科学基金（51976005）；国家科技重大专项（2017-Ⅱ-0005-0018）；先进航空动力创新工作站（依托中国航空发动机研究院设立）（HKCX2020-02-013）；航空发动机及燃气轮机基础科学中心项目（P2022-B-Ⅱ-004-001)

详细信息

作者简介:
龙嘉明（1997-），男，硕士生，研究领域为智能可控变形风扇技术。E-mail：zy2032119@buaa.edu.cn

通讯作者:
潘天宇（1989-），男，副教授，博士，研究领域为压气机稳定性和气动弹性。E-mail：pantianyu@buaa.edu.cn

中图分类号: V235.16
计量
- 文章访问数: 537
- HTML浏览量: 188
- PDF量: 90
- 被引次数: 0
出版历程
- 收稿日期: 2022-10-09
- 网络出版日期: 2023-05-29

Data-driven design method of controllable morphing blade profile

LONG Jiaming^1
,,
PAN Tianyu^{1, 2,*
, ,},
LI Chenzhang³,
ZHENG Mengzong¹,
LI Qiushi^{1, 4}

1.
Research Institute of Aero-Engine，Beihang University，Beijing 100191，China
2.
Advanced Jet Propulsion Creativity Center， Aero Engine Academy of China，Beijing 101300，China
3.
School of Energy and Power Engineering，Beihang University，Beijing 100191，China
4.
Key Laboratory of Fluid and Power Machinery，Xihua University，Chengdu 610039，China

摘要

摘要:
重点研究了综合考虑变形代价及气动收益的可控变形叶型优化设计方法。利用机器学习算法构建叶型几何与关键气动参数之间的预测模型，量化变形代价及气动收益，并搭建贝叶斯优化框架进行寻优。结果表明：基于机器学习的预测及优化框架能够准确预测风扇变形后的气动性能，且在考虑变形代价的条件下对叶型变形收益边界进行评估。主要结论是利用机器学习算法结合叶斯寻优框架可以获得兼顾变形代价以及气动收益的变形方案。相比于单纯的气动优化方案，此方案可以在保证气动性能提升的同时，使叶片最大应力降低14.17%，压电片驱动能耗降低67.45%。
- 超声风扇 /
- 可控变形 /
- 智能材料 /
- 机器学习 /
- 贝叶斯优化
Abstract:
The optimization design method of controllable morphing blade profile considering both morphing cost and aerodynamic benefits was discussed. A machine learning algorithm was used to build a prediction model to predict key aerodynamic parameters of morphing blade profiles. The morphing cost and aerodynamic benefits were quantified, and a Bayesian optimization framework was built for optimization. Results showed that the prediction and optimization framework based on machine learning can accurately predict the aerodynamic performance of the fan after morphing, and evaluate the profit boundary of the blade profile morphing considering the morphing cost. The main conclusion indicated that using machine learning algorithm and Bayesian optimization framework can obtain a morphing scheme taking into account both morphing cost and aerodynamic benefits. This scheme can reduce the maximum stress of blade by 14.17% and the energy consumption of piezoelectric actuator by 67.45% while ensuring the improvement of aerodynamic performance compared with the scheme only considering aerodynamic benefits.
- supersonic fan /
- controllable morphing /
- smart material /
- machine learning /
- Bayesian optimization

HTML

图 1 压电驱动变形风扇叶片示意图

Figure 1. Schematic diagram of piezoelectric driven deformed fan blade

下载: 全尺寸图片幻灯片

图 2 叶片有限元模型

Figure 2. Finite element model of blade

下载: 全尺寸图片幻灯片

图 3 计算网格

Figure 3. Computational grid

下载: 全尺寸图片幻灯片

图 4 Bowen悬臂梁模型示意图^[14]（单位：mm）

Figure 4. Schematic diagram of Bowen, cantilever beam model^[14] （unit：mm）

下载: 全尺寸图片幻灯片

图 5 有限元模型验证结果

Figure 5. Validation of the finite element method simulations

下载: 全尺寸图片幻灯片

图 6 ARL-SL19仿真及实验表面等熵马赫数分布对比

Figure 6. Comparison of simulated surface isentropic Mach number and experimental result of ARL-SL19

下载: 全尺寸图片幻灯片

图 7 叶片变形收益评估思路

Figure 7. Evaluation of blade deformation profit

下载: 全尺寸图片幻灯片

图 8 本文所用深度神经网络结构示意图

Figure 8. Structure schematic diagram of deep neural network used in this paper

下载: 全尺寸图片幻灯片

图 9 模型方均根误差随迭代步数变化

Figure 9. Variation of root mean square error of model with iteration steps

下载: 全尺寸图片幻灯片

图 10 气动参数预测结果与实际值对比

Figure 10. Comparison of aerodynamic between predicted results and actual result

下载: 全尺寸图片幻灯片

图 11 叶片表面应力分布

Figure 11. Stress distribution on blade surface

下载: 全尺寸图片幻灯片

图 12 寻优框架示意图

Figure 12. Schematic diagram of optimization framework

下载: 全尺寸图片幻灯片

图 13 本文叶型在不同来流工况下马赫数云图

Figure 13. Mach number contour of blade profile in this paper under different working conditions

下载: 全尺寸图片幻灯片

图 14 叶型总压比-流量特性线

Figure 14. Characteristic curves of total pressure ratio and mass flow of blade profiles

下载: 全尺寸图片幻灯片

图 15 叶型总压比-损失特性线

Figure 15. Characteristic curves of total pressure ratio and loss of blade profiles

下载: 全尺寸图片幻灯片

图 16 不同优化叶型与原始叶型几何对比

Figure 16. Geometric comparison between different optimized blade profiles and original blade profile

下载: 全尺寸图片幻灯片

图 17 不同优化叶型与原始叶型叶栅通道面积分布

Figure 17. Area distribution of cascade passage of different optimized blade profiles and original blade profile

下载: 全尺寸图片幻灯片

图 18 不同优化叶型在近堵点工况马赫数云图

Figure 18. Mach number contours at near choke condition of different optimized blade profiles

下载: 全尺寸图片幻灯片

图 19 优化叶型性能参数随权重系数ω₁变化

Figure 19. Variation of performance parameters of optimized blade profile with weight coefficient ω₁

下载: 全尺寸图片幻灯片

图 20 优化叶型气动参数随权重系数$ {\omega }_{2} $的变化示意图

Figure 20. Variation of aerodynamic parameters of optimized blade profile with weight coefficient $ {\omega }_{2} $

下载: 全尺寸图片幻灯片

图 21 优化叶型变形代价随权重系数$ {\omega }_{2} $的变化示意图

Figure 21. Variation of morphing cost of optimized blade profile with weight coefficient $ {\omega }_{2} $

下载: 全尺寸图片幻灯片

表 1 叶栅几何参数

Table 1. Cascade geometric parameters

参数	数值
叶栅稠度$ c $/s	1.866
安装角$\gamma / （{\text{°}}）$	124
最大相对厚度$ {t}_{\mathrm{m}\mathrm{a}\mathrm{x}} $/c	0.031
前缘相对厚度$ {r}_{\mathrm{L}\mathrm{E}} $/c	0.0024

下载: 导出CSV

表 2 模型预测精度

Table 2. Prediction accuracy of model

参数	$ {\delta }_{\mathrm{a}\mathrm{v}\mathrm{e}} $/%	${\delta }_{\mathrm{m}\mathrm{a}\mathrm{x} }/\text{%}$
$ {L}_{\mathrm{m}\mathrm{i}\mathrm{n}} $	1.416	2.037
$ {M}_{\mathrm{c}\mathrm{h}\mathrm{o}\mathrm{k}\mathrm{e}} $	0.208	0.515

下载: 导出CSV

表 3 部分权重系数下优化叶型气动性能

Table 3. Aerodynamic performance of optimized blade profile under some weight coefficient

案例	$ l $	${v}\mathrm{/V}$	$ {M}_{\mathrm{c}\mathrm{h}\mathrm{o}\mathrm{k}\mathrm{e}} $	$ {L}_{\mathrm{m}\mathrm{i}\mathrm{n}} $
原始叶型	0	0	1	0.0720
$ {\omega }_{1}=0 $	0.99	−3332	1.0183	0.0911
$ {\omega }_{1}=0.2 $	0.26	9960	1.0136	0.0445
$ {\omega }_{1}\gg 1 $	0.4	9427	0.9989	0.0411

下载: 导出CSV

表 4 部分权重系数下优化叶型气动性能及变形代价

Table 4. Aerodynamic parameters and morphing cost of optimized blade profile under some weight coefficient

案例	$ {L}_{\mathrm{m}\mathrm{i}\mathrm{n}} $	等效应力/MPa	$ {E}_{\mathrm{n}\mathrm{o}\mathrm{r}\mathrm{m}} $
原始叶型	0.0720	0	0
$ {\omega }_{2}=0 $	0.0411	642.8	0.9186
$ {\omega }_{2}=0.2 $	0.0422	551.7	0.299

下载: 导出CSV

参考文献(20)

[1]	PHAN T,SPRINGER P,LIEBICH R. Numerical investigation of an elastomer-piezo-adaptive blade for active flow control of a non-steady flow field using fluid-structure-interaction-simulations[J]. Journal of Turbomachinery,2017,139(9): 091004.1-091004.10.
[2]	KRONE J H,HUXDORF O,RIEMENSCHNEIDER J,et al. Experimental investigation and design of a shape-variable compressor cascade[J]. CEAS (Council of European Aerospace Societies) Aeronautical Journal,2017,8(1): 105-127.
[3]	MONNER H P, HUXDORF O, RIEMENSCHNEIDER J, et al. Design and manufacturing of morphing fan blades for experimental investigations in a cascaded wind tunnel[R]. Reston, US: the 23rd AIAA/AHS Adaptive Structures Conference, 2015.
[4]	LYU Z,MARTINS J R R A. Aerodynamic shape optimization of an adaptive morphing trailing edge wing[J]. Journal of Aircraft,2015,52(6): 1951-1970. doi: 10.2514/1.C033116
[5]	李刚. 含复铰可连续变弯度机翼机构设计与优化研究[D]. 哈尔滨: 哈尔滨工业大学, 2017. LI Gang. Design and optimization of camber continuously morphing wing mechanism with multiple joints[D]. Harbin: Harbin Institute of Technology, 2017. (in Chinese)
[6]	王晓明. 压电驱动柔性翼面的优化册与变形控制方法[D]. 辽宁大连: 大连理工大学, 2018. WANG Xiaoming. Optimal design and shape control method for piezo-actuated flexible wing surface[D]. Dalian Liaoning: Dalian University of Technology, 2018. (in Chinese)
[7]	桂幸民, 滕金芳, 刘宝杰. 航空压气机气动热力学理论与应用[M]. 上海: 上海交通大学出版社, 2014.
[8]	彭泽琰, 刘刚. 航空燃气轮机原理[M]. 北京: 国防工业出版社, 2000: 3-4.
[9]	石超. 超音速压气机预压缩叶型设计方法及研究[D]哈尔滨: 哈尔滨工业大学, 2018. SHI Chao. Study on design method for pre-compression blade profile of supersonic compressor[D]. Harbin: Harbin Institute of Technology, 2018. (in Chinese)
[10]	KANTROWITZ A. The supersonic axial-flow compressor[R]. NACA-Report 974/NACA-ACR- L6D02, 1950.
[11]	STEVEN L,BRUNTON J,KUTZ N,et al. Data-driven aerospace engineering: reframing the industry with machine learning[J]. AIAA Journal,2021,59(8): 2820-2847.
[12]	汪逸然,赵文军,梁连国,等. 基于机器学习方法的流体机械气动优化设计研究现状及展望[J]. 风机技术,2020,62(5): 77-90. doi: 10.16492/j.fjjs.2020.05.0011 WANG Yiran,ZHAO Wenjun,LIANG Lianguo,et al. A review on optimal design of fluid machinery based on machine learning method[J]. Chinese Journal of Turbomachinery,2020,62(5): 77-90. (in Chinese) doi: 10.16492/j.fjjs.2020.05.0011
[13]	张学伍,曾涛,赵程,等. 宏观纤维复合材料及其驱动器的研究与应用[J]. 机械工程材料,2020,44(6): 77-81. doi: 10.11973/jxgccl202006017 ZHANG Xuewu,ZENG Tao,ZHAO Cheng,et al. Research and application of macro fiber composite and actuator[J]. Materials for Mechanical Engineering,2020,44(6): 77-81. (in Chinese) doi: 10.11973/jxgccl202006017
[14]	BOWEN C R,GIDDINGS P F,SALO A I T,et al. Modeling and characterization of piezoelectrically actuated bistable composites[J]. IEEE transactions on Ultrasonics, Ferroelectrics, and Frequency Control,2011,58(9): 1737-1750. doi: 10.1109/TUFFC.2011.2011
[15]	TWEEDT D L,SCHREIBER H A,STARKEN H. Experimental investigation of the performance of a supersonic compressor cascade[J]. Journal of Turbomachinery,1988,110(4): 456-466. doi: 10.1115/1.3262219
[16]	SEROVY G K,OKIISHI T H. Performance of a compressor cascade configuration with supersonic entrance flow: a review and comparison of experiments in three installations[J]. Journal of Turbomachinery,1988,110(4): 441-449. doi: 10.1115/1.3262217
[17]	BRUNTON S L,NOACK B R,KOUMOUTSAKOS P. Machine learning for fluid mechanics[J]. Annual Review of Fluid Mechanics,2020,52: 477-508. doi: 10.1146/annurev-fluid-010719-060122
[18]	SUN Dongchang,TONG Liyong. Design optimization of piezoelectric actuator patterns for static shape control of smart plates[J]. Smart Materials and Structures,2005,14(6): 1353-1362. doi: 10.1088/0964-1726/14/6/027
[19]	KUCUK I,YILDIRIM K,SADEK I,et al. Optimal control of a beam with Kelvin-Voigt damping subject to forced vibrations using a piezoelectric patch actuator[J]. Journal of Vibration and Control,2015,21(4): 701-713. doi: 10.1177/1077546313488617
[20]	ZAVADSKAS E K,TURSKIS Z,DEJUS T,et al. Sensitivity analysis of a simple additive weight method[J]. International Journal of Management and Decision Making,2007,8(5/6): 555-574. doi: 10.1504/IJMDM.2007.013418