English

应用科学学报 ›› 2026, Vol. 44 ›› Issue (3): 390-408.doi: 10.3969/j.issn.0255-8297.2026.03.004

• 智能信息处理 • 上一篇    

基于AttResVGG模型的燃烧室点火预测算法

廖清1, 陈泓佑1, 刘重阳2, 屈凌峰3, 段纪平2, 夏萍1, 田宝单4, 范勇1   

  1. 1. 西南科技大学四川省大数据与智能系统工程技术研究中心, 四川 绵阳 621010;
    2. 中国航发四川燃气涡轮研究院燃烧试验研究部, 四川 绵阳 621050;
    3. 广州大学网络空间先进技术研究院, 广东 广州 510006;
    4. 西南科技大学数理学院, 四川 绵阳 621010
  • 收稿日期:2025-08-27 发布日期:2026-06-23
  • 通信作者: 陈泓佑，博士，讲师，研究方向为深度学习、图像处理、数据科学。E-mail:chy2019@foxmail.com E-mail:chy2019@foxmail.com
  • 基金资助:
    四川省自然科学基金重大项目(No.2025ZNSFSC0005);国家自然科学基金项目(No.62402125)

Combustion Chamber Ignition Prediction Algorithm Based on AttResVGG Model

LIAO Qing1, CHEN Hongyou1, LIU Chongyang2, QU Lingfeng3, DUAN Jiping2, XIA Ping1, TIAN Baodan4, FAN Yong1   

  1. 1. Sichuan Big Data and Intelligent System Engineering Technology Research Center, Southwest University of Science and Technology, Mianyang 621010, Sichuan, China;
    2. Combustion Testing and Research Department, AECC Sichuan Gas Turbine Establishment, Mianyang 621050, Sichuan, China;
    3. Cyberspace Institute of Advanced Technology, Guangzhou University, Guangzhou 510006, Guangdong, China;
    4. School of Mathematics and Physics, Southwest University of Science and Technology, Mianyang 621010, Sichuan, China
  • Received:2025-08-27 Published:2026-06-23

PDF

7

摘要： 针对航空发动机燃烧室点火预测中工况特征复杂，训练样本量小且数据不平衡，一般深度学习预测模型难以满足预测任务需求的问题，提出了融合残差连接与自注意力机制的网络模型——注意力残差视觉几何组网络（attention residual visual geometry group,Att ResVGG）。该模型通过多头自注意力机制捕捉工况参数间的依赖关系，建立工况参数与点火状态之间的映射关系。针对数据量不足和类别不平衡问题，设计了基于物理约束的数据增广策略，通过维持燃油空气比、温度压强比等关键物理参数关系合成新工况样本。此外，设计基于贝叶斯优化自动机器学习算法优化模型超参数，进一步提升模型预测能力。为验证模型效果，在两个数据集上的实验表明，AttResVGG模型的准确率分别达97.67%和90.48%,Kappa系数分别达0.950 5和0.834 6，优于所对比的模型。

关键词: 深度学习, 点火预测, 数据增广, 自注意力机制, 自动机器学习

Abstract: To address the challenges posed by complex operating conditions, limited training samples, data imbalance, and the inability of conventional deep learning models to meet the requirements of ignition prediction for aero-engine combustion chambers, a network model integrating residual connections and a self-attention mechanism, called the attention residual visual geometry group network(AttResVGG), was proposed. The model used a multi-head self-attention mechanism to capture dependencies among operating condition parameters and establish a mapping between these parameters and ignition status. To address insufficient data size and class imbalance, a physically constrained data augmentation strategy was designed to synthesize new operating condition samples while maintaining key physical parameter relationships, such as the fuel-air ratio and temperature-pressure ratio.In addition, an automated machine learning algorithm based on Bayesian optimization was designed to optimize model hyperparameters, further enhancing the model's predictive performance. To validate the effectiveness of this model, experiments on two datasets show that the accuracy of the AttResVGG model reaches 97.67% and 90.48%, and the Kappa coefficients reach 0.950 5 and 0.834 6, respectively, which are better than those of the compared models.

Key words: deep learning, ignition prediction, data augmentation, self-attention mechanism, automated machine learning

中图分类号: