A Generalized Combustion Characteristic Prediction Model Based on Flame Hyperspectral Technology
WANG Ming1, 2, HE Hong-juan1, 2, WANG Bao-rui1, 2*, AI Yu-hua1, 2, WANG Yue1, 2
1. New Technology Laboratory, Institute of Engineering Thermophysics, Chinese Academy of Sciences, Beijing 100190, China
2. University of Chinese Academy of Sciences, Beijing 100049, China
摘要： 燃烧特性的准确预测可以为燃烧技术的优化提供重要依据。利用甲烷火焰高光谱数据同时预测多种燃烧特性，以解决工程中监测燃烧特性需要多种采集设备及现场取样的难题，提出了一种基于三维卷积神经网络(3-dimensional Convolutional Neural Network)方法的火焰特性预测模型。首先在层流火焰高光谱数据上验证了该方法并进一步将模型应用于湍流火焰高光谱数据，接着建立了一种通用模型和训练方法，最后使用Gradient-weighted Class Activation Mapping(Grad-CAM)方法对模型进行可解释性研究。在层流模型中研究了不同参数对火焰特性预测模型准确性的影响；在湍流模型中探究了使用迁移学习的可行性以及添加小波去噪处理方法和改变卷积层数对模型结果的影响；在通用模型中解决数据不一致的问题，使得层流和湍流火焰作为训练数据对模型进行混合训练成为可能，实现了从实验室小火焰到实际工程火焰的燃烧特性参数预测模型的建立；在模型的可解释性研究中使用Grad-CAM方法对神经网络模型的工作原理和决策过程进行研究。结果显示层流预测模型中针对不同预测参数平均准确率均能达到96%以上，使用Mish+Mae的参数组合平均准确率最高为98%且拥有较高的模型稳定性；湍流预测模型的平均准确率均能达到95%以上，迁移学习、小波去噪处理和改变卷积层数均能在一定程度上提升模型性能，但需要进行合适的调整；通用模型在层流和湍流不同参数预测的平均准确率分别达到了97%和94%，在两种不同火焰的工况下均达到较高的准确率，说明模型的适用性较强；可解释性研究中Grad-CAM热力图上的信息表明该方法能够通过定位火焰敏感区域，并从光谱中提取到关键组分信息，从而完成对火焰特性的预测。该研究可以为燃烧学实验测试技术的智能化发展提供支持，对燃烧特性的研究具有重要意义。
关键词：高光谱技术；燃烧特性；神经网络
Abstract：Accurately measuring combustion characteristics can provide important references for optimizing combustion technology. This study predicts multiple combustion properties simultaneously using high spectral images of methane flames, addressing the challenge of monitoring combustion characteristics in engineering, which typically requires various collection devices and on-site sampling. A flame characteristic prediction model based on the 3D-CNN method is proposed. Initially, this method is validated on laminar flame high spectral images and further applied to turbulent flame high spectral images. Subsequently, a universal model and training method are established. Finally, the Grad-CAM method is used to conduct interpretability studies of the model. In the laminar flow model, we investigated the influence of different parameters on the accuracy of flame characteristic prediction models. In the turbulent flow model, we explored the feasibility of using transfer learning and the impact of adding wavelet denoising techniques and altering the number of convolutional layers on the model results. Within the general model, we addressed the issue of inconsistent labels in different datasets, enabling the blending of laminar and turbulent flame data for mixed training of the model. This facilitated the establishment of a combustion characteristic prediction model, ranging from laboratory-scale flames to real-world engineering flames. In the study of model interpretability, the Grad-CAM method was employed to investigate the operational principles and decision-making processes of neural network models. The results indicate that the average accuracy of the laminar flow prediction model for different predictive parameters exceeds 96%, with the Mish+Mae parameter combination achieving the highest average accuracy of 98% and demonstrating enhanced model stability. The average accuracy in the turbulent flow prediction model is consistently above 95%. Transfer learning, wavelet denoising, and altering the number of convolutional layers all contribute to improving model performance to some extent, contingent upon appropriate adjustments. The general model achieves an average of 97% and 94% accuracy for laminar and turbulent parameter predictions, respectively, while exhibiting higher robustness. In the interpretability study, information from Grad-CAM heatmaps suggests that this method can effectively identify flame-sensitive regions, extract crucial component information from spectra, and accomplish flame characteristic predictions. This study supports the intelligent development of experimental testing techniques in combustion science, holding significant implications for investigating combustion characteristics.
汪 明，贺红娟，王宝瑞，艾育华，王 岳. 一种通用的基于火焰高光谱的燃烧特性预测模型[J]. 光谱学与光谱分析, 2025, 45(02): 532-541.
WANG Ming, HE Hong-juan, WANG Bao-rui, AI Yu-hua, WANG Yue. A Generalized Combustion Characteristic Prediction Model Based on Flame Hyperspectral Technology. SPECTROSCOPY AND SPECTRAL ANALYSIS, 2025, 45(02): 532-541.
[1] LUO Jing, MA Bai-tan, KANG Zhe(雒 晶，马百坦，康 哲). Proceedings of China Society of Automotive Engineers Annual Conference and Exhibition(中国汽车工程学会暨展览会论文集), 2022, 1044.
[2] Gao Y, Liu Y, Dong Z, et al. Energy, 2023, 268: 126741.
[3] Wang Q, Mei X H, Wu Y, et al. Flow, Turbulence and Combustion, 2021，107: 609.
[4] Susa A J, Zheng L, Hanson R K. Schlieren-Based Measurements of Propane Flame Speeds at Extreme Temperatures, 12th US National Combustion Meeting，2021.
[5] Liu N, Chen T Y, Zhong H, et al. Applied Physics Letters, 2022, 120(20): 201103.
[6] Rubiella C, Byun H, Park Y, et al. Journal of Engineering for Gas Turbines and Power, 2024, 146(9).
[7] SU Hong-jun（苏红军）. National Remote Sensing Bulletin（遥感学报），2022，26(8): 1504.
[8] Si M, Cheng Q, Yuan L, et al. Combustion and Flame, 2021, 233: 111568.
[9] WANG Yan, WANG Bao-rui, WANG Yue（王 岩, 王宝瑞, 王 岳）. Spectroscopy and Spectral Analysis（光谱学与光谱分析）, 2022, 42(8): 2403.
[10] Qiu Xuanbing, Xi Tingyu, Sun Dongyuan, et al. Fire Technology, 2018, 54: 1249.
[11] Manolakis D G, Marden D, Kerekes J P, et al. Algorithms for Multispectral, Hyperspectral, and Ultraspectral Imagery Ⅶ, SPIE, 2001, 4381: 308(https://doi.org/10.1117/12.437021).
[12] Plaza A, Benediktsson J A, Boardman J W, et al. Remote Sensing of Environment, 2009, 113: S110.
[13] Paoletti M E, Haut J M, Plaza J, et al. ISPRS Journal of Photogrammetry and Remote Sensing, 2019, 158: 279.
[14] SHE Hai-long, XIE Shan-juan, ZOU Jing-jie, et al(佘海龙，解山娟，邹静洁). Computer Engineering and Applications(计算机工程与应用), 2021, 57(4): 169.
[15] Wang Y, Zhao W J, Wang B R, et al. Applied Optics, 2022, 61(7): D75.
[16] Selvaraju R R, Cogswell M, Das A, et al. Grad-CAM: Visual Explanations From Deep Networks Via Gradient-Based Localization. In Proceedings of the IEEE International Conference on Computer Vision, 2017.
[17] XIE Jie-cheng, ZHANG Da-li, XU Wen-li（谢杰成, 张大力, 徐文立）. Journal of Image and Graphics（中国图象图形学报）, 2002, 7(3): 209.
[18] Misra D. A Self Regularized Non-Monotonic Activation Function, 2019, arXiv: 1908. 08681.
[19] Golik P, Doetsch P, Ney H, et al. Interspeech, 2013, 13: 1756.
