面向光场火焰温度场重建的特征提取优化方法

doi:10.16085/j.issn.1000-6613.2023-1446

摘要/Abstract

摘要：

近年来，利用深度学习的方法从光场火焰图像中重建三维温度分布已成为辐射测温研究领域的一个新方向。传统的温度场重建网络仍沿用平面图像的特征提取方法，不仅忽略了光场图像中记录的三维光线信息，而且没有充分考虑追迹光线的分类问题，致使先验信息的缺失和不同类型特征的杂糅，对温度场的重建精度产生负面影响。本文对网络中的特征提取过程进行了优化，首先将子孔径图像的视角信息添加到网络输入端，然后使用双分支卷积方法分别提取光场的空间特征和角度特征，最后利用注意力机制对特征在不同尺度上的重要性进行建模。本文通过正交实验验证了上述因素对重建精度的影响。仿真结果表明，与传统网络相比，优化方法重建的温度场平均相对误差降低了44.82%，最大相对误差降低了34.76%。

关键词: 温度场重建, 优化, 神经网络, 数值分析

Abstract:

In recent years, the reconstruction of the three-dimensional (3D) temperature field from the flame light-field image by the method of deep learning has been a new direction in radiation thermometry. The traditional temperature field reconstruction network still uses the feature extraction method of the plane image, which not only ignores the 3D ray information recorded in the light-field image but also fails to consider the classification of tracing rays. Therefore, the absence of prior information and the hybridity of different types of features have a negative impact on the reconstruction accuracy of temperature field. This article optimized the feature extraction process in the network. Firstly, the view angle information of the sub-aperture image was added to the network input. Then the spatial and angular features of the light-field were extracted by the double-branch convolution method. Finally, the attention mechanism was used to model the importance of features at different scales. The effectiveness of the above factors on the reconstruction accuracy was verified by orthogonal experiments. The simulation results showed that the Mean Relative Error (MRE) of the temperature field reconstructed by the optimization method was reduced by 44.82%, and the Maximum Relative Error (MMRE) was reduced by 34.76% compared to traditional networks.

Key words: temperature field reconstruction, optimization, neural networks, numerical analysis

中图分类号:

TK311

单良, 华夏杰, 牛玉风, 赵腾飞, 洪波, 孔明. 面向光场火焰温度场重建的特征提取优化方法[J]. 化工进展, 2024, 43(2): 823-829.

SHAN Liang, HUA Xiajie, NIU Yufeng, ZHAO Tengfei, HONG Bo, KONG Ming. Optimization method for light-field feature extraction in flame temperature field reconstruction[J]. Chemical Industry and Engineering Progress, 2024, 43(2): 823-829.

图/表 14

参考文献 27

1	SHARMA Shirish Kumar, SHIVAPUJI Anand M, DASAPPA S. Characterization of a novel two-stage high ash coal gasifier for low tar and high calorific value gas[J]. Applied Thermal Engineering, 2024, 237: 121775.
2	BOTTA Luigi, LA MANTIA Francesco Paolo, CERAULO Manuela, et al. Effect of processing temperature and mixing time on the properties of PP/GnP nanocomposites[J]. Polymer Degradation and Stability, 2020, 181: 109321.
3	BOOMADEVI P, PAULSON V, SAMLAL Stanley, et al. Impact of microalgae biofuel on microgas turbine aviation engine: A combustion and emission study[J]. Fuel, 2021, 302: 121155.
4	HWANG Ouk, LEE Min Chul, WENG Wubin, et al. Development of novel ultrasonic temperature measurement technology for combustion gas as a potential indicator of combustion instability diagnostics[J]. Applied Thermal Engineering, 2019, 159: 113905.
5	SHCHEPAKINA Elena Anatolievna, ZUBRILIN Ivan Alexandrovich, KUZNETSOV Alexey Yurievich, et al. Physical and chemical features of hydrogen combustion and their influence on the characteristics of gas turbine combustion chambers[J]. Applied Sciences, 2023, 13(6): 3754.
6	SALINAS Carlos T, PU Yang, LOU Chun, et al. Experiments for combustion temperature measurements in a sugarcane bagasse large-scale boiler furnace[J]. Applied Thermal Engineering, 2020, 175: 115433.
7	ELIAS Jessy, FACCINETTO Alessandro, BATUT Sebastien, et al. Thermocouple-based thermometry for laminar sooting flames: Implementation of a fast and simple methodology[J]. International Journal of Thermal Sciences, 2023, 184: 107973.
8	WU Wendong, ADEOSUN Adewale, AXELBAUM Richard L. A new method of flame temperature measurement utilizing the acoustic emissions from laser-induced plasmas[J]. Proceedings of the Combustion Institute, 2019, 37(2): 1409-1415.
9	LIU Hecong, SUN Bin, CAI Weiwei. kHz-rate volumetric flame imaging using a single camera[J]. Optics Communications, 2019, 437: 33-43.
10	KIM Dennis K, SUNDERLAND Peter B. Fire ember pyrometry using a color camera[J]. Fire Safety Journal, 2019, 106: 88-93.
11	QI Qi, HOSSAIN Md Moinul, LEI Gang, et al. Optimum angular arrangement of a multi-light field imaging technique for flame temperature reconstruction[J]. Measurement, 2022, 204: 112110.
12	ZHANG Biao, WANG Chen, LIU Yudong, et al. Reconstruction of 3D temperature profile of radiative participatory flame based on digital refocusing technique of light field camera[J]. International Journal of Photoenergy, 2019, 2019: 1-13.
13	RODRÍGUEZ A, ESCUDERO F, CRUZ J J, et al. Retrieving soot volume fraction fields for laminar axisymmetric diffusion flames using convolutional neural networks[J]. Fuel, 2021, 285: 119011.
14	唐广通, 许烨烽, 闫慧博, 等. 基于深度学习与热辐射成像耦合的炉内温度场在线测量[J]. 动力工程学报, 2022, 42(10): 960-966.
	TANG Guangtong, XU Yefeng, YAN Huibo, et al. Research of on-line measurement of temperature field in furnaces based on deep learning coupled thermal radiative imaging[J]. Journal of Chinese Society of Power Engineering, 2022, 42(10): 960-966.
15	李智聪, 娄春. 基于深度学习的乙烯层流扩散火焰温度和碳烟体积分数分布重建[J]. 燃烧科学与技术, 2022, 28(2): 198-205.
	LI Zhicong, LOU Chun. Reconstruction of temperature and soot volume fraction distribution of ethylene laminar diffusion flame based on deep learning[J]. Journal of Combustion Science and Technology, 2022, 28(2): 198-205.
16	QU Xiangju, SONG Yang, JIN Ying, et al. 3D particle field reconstruction method based on convolutional neural network for SAPIV[J]. Optics Express, 2019, 27(8): 11413.
17	WANG Zhenyu, SONG Chunfeng, CHEN Tao. Deep learning based monitoring of furnace combustion state and measurement of heat release rate[J]. Energy, 2017, 131: 106-112.
18	JIN Ying, ZHANG Wanqing, SONG Yang, et al. Three-dimensional rapid flame chemiluminescence tomography via deep learning[J]. Optics Express, 2019, 27(19): 27308-27334.
19	张杰, 齐琪, 韩哲哲, 等. 基于深度学习和光场成像的火焰三维温度场重建算法[J]. 东南大学学报(自然科学版), 2021, 51(6): 1060-1067.
	ZHANG Jie, QI Qi, HAN Zhezhe, et al. Reconstruction algorithm of flame 3D temperature distribution based on deep learning and light field imaging[J]. Journal of Southeast University (Natural Science Edition), 2021, 51(6): 1060-1067.
20	LI Jian, HOSSAIN Md Moinul, SUN Jun, et al. Simultaneous measurement of flame temperature and absorption coefficient through LMBC-NNLS and plenoptic imaging techniques[J]. Applied Thermal Engineering, 2019, 154: 711-725.
21	TORU Iwane, MARIE Shoda. Light-field camera and display as information retrieval systems for three-dimensional images[J]. Optical Engineering, 2018, 57(6): 061616.
22	WAFA Abrar, POURAZAD Mahsa T, NASIOPOULOS Panos. A deep learning based spatial super-resolution approach for light field content[J]. IEEE Access, 2020, 9: 2080-2092.
23	WANG Xinya, MA Jiayi, GAO Wenjing, et al. MPIN: A macro-pixel integration network for light field super-resolution[J]. Frontiers of Information Technology & Electronic Engineering, 2021, 22(10): 1299-1310.
24	YUE Huanjing, SHEN Sheng, YANG Jingyu, et al. Reference image guided super-resolution via progressive channel attention networks[J]. Journal of Computer Science and Technology, 2020, 35(3): 551-563.
25	LI Fengpeng, FENG Ruyi, HAN Wei, et al. Ensemble model with cascade attention mechanism for high-resolution remote sensing image scene classification[J]. Optics Express, 2020, 28(15): 22358.
26	Sanghyun WOO, PARK Jongchan, LEE Joon-Young, et al. CBAM: Convolutional block attention module[C]//Computer Vision - ECCV 2018: 15th European Conference, Munich, Germany, September 8-14, 2018, Proceedings, Part Ⅶ. New York: ACM, 2018: 3-19.
27	INAN Onur, UZER Mustafa Serter. A method of classification performance improvement via a strategy of clustering-based data elimination integrated with K-fold cross-validation[J]. Arabian Journal for Science and Engineering, 2021, 46(2): 1199-1212.

试验号	因素
试验号	A(图像格式)	B(CBAM)	C(HFEM)
1	1(MPI)	1(不采用)	1(不采用)
2	1	1	2(采用)
3	1	2(采用)	1
4	1	2	2
5	2(SAI)	1	1
6	2	1	2
7	2	2	1
8	2	2	2

试验号	因素
试验号	A(图像格式)	B(CBAM)	C(HFEM)
1	1(MPI)	1(不采用)	1(不采用)
2	1	1	2(采用)
3	1	2(采用)	1
4	1	2	2
5	2(SAI)	1	1
6	2	1	2
7	2	2	1
8	2	2	2

试验号	因素							指标
试验号	A	B	A×B	C	C×A	B×C	误差项	MRE/%	MMRE/%
1	1	1	1	1	1	1	1	0.288	3.513
2	1	1	1	2	2	2	2	0.278	3.645
3	1	2	2	1	1	2	2	0.387	2.858
4	1	2	2	2	2	1	1	0.343	2.794
5	2	1	2	1	2	1	2	0.620	3.721
6	2	1	2	2	1	2	1	0.487	3.947
7	2	2	1	1	2	2	1	0.250	3.987
8	2	2	1	2	1	1	2	0.155	2.287
MRE K_j₁	1.296	1.674	0.971	1.545	1.317	1.406	1.368
MRE K_j₂	1.512	1.135	1.837	1.264	1.491	1.402	1.440
MMRE K_j₁	12.810	14.826	13.431	14.078	12.605	12.314	14.241
MMRE K_j₂	13.941	11.925	13.320	12.673	14.146	14.437	12.510

试验号	因素							指标
试验号	A	B	A×B	C	C×A	B×C	误差项	MRE/%	MMRE/%
1	1	1	1	1	1	1	1	0.288	3.513
2	1	1	1	2	2	2	2	0.278	3.645
3	1	2	2	1	1	2	2	0.387	2.858
4	1	2	2	2	2	1	1	0.343	2.794
5	2	1	2	1	2	1	2	0.620	3.721
6	2	1	2	2	1	2	1	0.487	3.947
7	2	2	1	1	2	2	1	0.250	3.987
8	2	2	1	2	1	1	2	0.155	2.287
MRE K_j₁	1.296	1.674	0.971	1.545	1.317	1.406	1.368
MRE K_j₂	1.512	1.135	1.837	1.264	1.491	1.402	1.440
MMRE K_j₁	12.810	14.826	13.431	14.078	12.605	12.314	14.241
MMRE K_j₂	13.941	11.925	13.320	12.673	14.146	14.437	12.510

源	MRE显著性	MMRE显著性
图像格式 (A)	0.204	0.631
CBAM (B)	0.160	0.566
HFEM (C)	0.085	0.342
A×B	0.053	0.959
B×C	0.249	0.537
A×C	0.964	0.435