引入物理约束的航空发动机燃烧室温度场预测模型
链接：
https://mp.weixinbridge.com/mp/wapredirect?url=https%3A%2F%2Fbook.yunzhan365.com%2Fijzxj%2Fhcks%2Fmobile%2Findex.html&action=appmsg_redirect&uin=MTI4OTExNzQ4NA%3D%3D&biz=MzI1NzA3ODU1OA==&mid=2686279232&idx=1&type=1&scene=0
引入注意力模块双支路网络架构
本文引入物理约束的损失函数架构
研究背景
未来高性能飞行器的实现对航空发动机的推力提出新的要求，而推力的增加势必要求发动机燃烧室更高的温升能力。受限于恶劣的工作环境，航空发动机燃烧室往往承受着较高的热载荷。而燃烧室温升的进一步增加无疑为发动机的寿命和工作性能带来了新的矛盾。因此，为了在现有材料结构的基础上使燃烧室涡轮前温度最大化，需要对燃烧室的温度分布进行一定程度的调节，避免局部超温现象的发生。美国曾提出通过综合高性能涡轮发动机技术计划（IHPTET）减小出口温度场周向分布系数（OTDF）来提高燃烧室的出口温度。为了实现这一控制过程，需要建立一个较为准确的燃烧室区域温度分布模型。然而，航空发动机燃烧过程中往往伴随着复杂的燃烧现象。在这样复杂的燃烧环境下，准确预测燃烧室的流量参数分布，特别是燃烧室内的温度分布，对燃烧室的设计、运行乃至优化都有着非常重要的作用。
解决的问题及创新点
针对发动机燃烧室内温度场的宽范围、高精度预测问题，本文构建了一种具有高精度的注意力双支路网络预测模型，并基于这一网络框架，引入截面平均温度偏差的损失函数项进行辅助训练，所得到的注意力网络模型相较于传统的卷积网络和全连接网络在学习区间内外都有着更高的预测精度。
总结与展望
本文构建了一种具有高精度的注意力双支路网络模型，并引入截面平均温度偏差的物理损失函数对网络模型学习进行优化，所得到的网络模型在学习区间内外都可以保持较高的预测精度。通过均方差和物理约束组合训练的方式，网络模型可以在保证精度的同时更贴近于我们所给出的物理约束，从而增强网络模型在更宽范围的预测能力，本文对宽范围内温度场网络预测模型的优化方法可以为后续深度学习在流场预测中遇到的可解释性问题和泛化能力问题提供参考。
近几年论文
[1]Ziao Wang, Renzhe Huang, Yiming Li, Jialin Zheng, Jifeng Guo, and Juntao Chang*. Experimental and numerical simulation of shock train characteristics in an isolator with incident shocks. Aerospace Science and Technology, 2023: 108309.
[2]Chengkun Lv, Qian Huang, Juntao Chang*, Ziao Wang, Jialin Zheng, and Daren Yu. Mode transition path optimization for turbine-based combined-cycle ramjet stage under uncertainty propagation of integrated airframe-propulsion system. Energy, 2023: 126718.
[3]Xuan Wang, Chen Kong, Minghao Ren, Aihan Li, Juntao Chang*. Research on temperature field prediction method in an aero-engine combustor with high generalization ability. Applied Thermal Engineering, Volume 239,2024,122042,
[4]Ziao Wang, Juntao Chang*, Chen Kong, Renzhe Huang, and Xuanan Xin. Experimental investigation of micro-ramp control for shock train under various incoming flow conditions. Physical Review Fluids, 2022, 7(10): 103401.
[5]Yunfei Li, Chengkun Lv, Juntao Chang*, Ziao Wang, and Chen Kong. A Bayesian data assimilation method to enhance the time sequence prediction ability of data-driven models. AIP Advances, 2022, 12(10): 105021.
[6]Chengkun Lv, Haiqi Xu, Juntao Chang*, Youyin Wang, Ruoyu Chen, and Daren Yu. Mode transition analysis of a turbine-based combined-cycle considering ammonia injection pre-compressor cooling and variable-geometry ram-combustor. Energy, 2022, 261: 125324.
[7]Ziao Wang, Juntao Chang*, Yiming Li, Ruoyu Chen, Wenxin Hou, Jifeng Guo and Lianjie Yue. Oscillation of the shock train under synchronous variation of incoming Mach number and backpressure. Physics of Fluids, 2022, 34(4): 046104.
[8]Ziao Wang, Kejing Xu and Juntao Chang*. Distributed Fluidic Control Method for Alleviating Rapid Movement of Shock Train. AIAA Journal, 2022: 1-18.
[9]Chen Kong, Fuxu Quan, Yunfei Li, Jingfeng Tang and Juntao Chang*. Prediction model of temperature field in dual-mode combustors based on wall pressure. Acta Astronautica, 2022, 196: 73-84.
[10]Ziao Wang, Xuanan Xin, Renzhe Huang, Chen Kong, Chengkun Lv and Juntao Chang*. Mechanism of shock-train/boundary-layer interaction in spanwise concave isolator channels. Acta Astronautica, 2022, 199: 232-248.
[11]Chengkun Lv, Haiqi Xu, Fuxu Quan, Juntao Chang*, and Daren Yu. Thermodynamic modeling and analysis of ammonia injection pre-compressor cooling cycle: A novel scheme for high Mach number turbine engines. Energy Conversion and Management, 2022, 265: 115776.
[12]Chen Kong, Ziao Wang, Junlong Zhang, Xuan Wang, Kai Wang, Yunfei Li and Juntao Chang*. Research on flame prediction in a scramjet combustor using data-driven model. Physics of Fluids, 2022, 34: 066101.
[13]Chengkun Lv, Juntao Chang*, Wen Bao, and Daren Yu. Recent research progress on airbreathing aero-engine control algorithm. Propulsion and Power Research, 2022, 11(1): 1-57.
[14]Yunfei Li, Ziao Wang*, Weiyu Jiang, Zongqi Xie, Chen Kong, and Juntao Chang*. Research on time sequence prediction of the flow field structure of supersonic cascade channels in wide range based on artificial neural network. Physics of Fluids, 2022, 34: 016106.
[15]Chen Kong, Chenlin Zhang, Ziao Wang, Yunfei Li and Juntao Chang*. Efficient Prediction of Supersonic Flowfield in an Isolator Based on Pressure Sequence. AIAA Journal, 2022, 60(5): 2826-2835.
[16]Yunfei Li, Juntao Chang*, Chen Kong, and Wen Bao. Recent progress of machine learning in flow modeling and active flow control. Chinese Journal of Aeronautics, 2022, 35(4): 14-44.
团队介绍
常军涛（通讯作者），国家杰青，哈尔滨工业大学教授，长期从事高超声速推进系统方面的科研教学工作。发展了高超声速进气道不起动预警、监测和控制方法，建立了高超声速发动机多模式协调控制方法。任中国空天动力联合会发动机控制组专业委员会主任、中国空气动力学学会理事、中国工程热物理学会热机气动热力学分会委员、中国力学学会第10届激波与激波管专业委员会委员。任《推进技术》副主编、《气动研究与试验》副主编。获得省部级一等奖、二等奖共计3项科研奖励，在JFM、AIAA、POF等国际著名期刊发表SCI检索论文100余篇，授权发明专利20余项，出版学术专著2部。2021年获国家杰出青年基金项目；2020年入选中国高被引学者榜单（航天工程学科）；2019年获工程热物理领域吴仲华优秀青年学者奖；2017年获国家优秀青年基金项目，获冲压发动机领域兴洲奖，获国防科学技术发明二等奖，排序第一。