Flow pattern identification model of gas-oil two-phase flow in the scavenge pipe with images processing
-
摘要:
为探究航空发动机回油管尺度下以润滑油和封严空气为主要介质的油气两相流相分布特征，实现回油管内流型准确辨识，提出了一种基于图像处理的两相流型辨识模型。该模型基于航空发动机工况下水平回油管内油气两相流实验获取的4种典型流型图像结果，以双边滤波、二值化、小波分解等图像处理技术提取的流型特征数据集为输入，采用Elman循环神经网络模型实现流型辨识。结果表明：在航空发动机工况范围内，回油管内会出现4种典型流型：弹状流、层状流、波状流和环状流。基于流型图像的辨识模型能够成功辨识4种不同流型，且在验证集上辨识准确率为93.06%、宏F1为97.60%，具有较好的准确性和稳健性。
-
关键词:
- 水平回油管 /
- 两相流型辨识 /
- 图像处理 /
- 小波分解 /
- Elman循环神经网络
Abstract:To explore the phase distribution characteristics of the oil-gas two-phase flow in scavenge pipe of aero-engine lubrication system, and achieve accurate flow pattern identification, a flow pattern identification model with image processing was proposed based on the typical flow patterns images obtained from a horizontal pipe under the working condition of aero-engine. Four typical flow patterns emerged in this experiment: slug, stratified, wavy, and annular flow. By image processing technologies such as bilateral filter, binarization, and wavelet decomposition, feature parameters were extracted from the images and used as input to the model. The identification model based on Elman Recurrent Neural Networks was established through training and verification, and it can successfully identify four different flow patterns. The identification accuracy of the model was 93.06%, and the robustness index macro-F1 was 97.60% on the verification set.
-
表 1 滑油和空气物性（20 ℃）
Table 1. Properties of oil and air （20 ℃）
介质 型号 密度/
（kg/m3）动力黏度/
（mPa·s）润滑油 飞马Ⅱ号 998.5 60 空气 1.205 0.018 
下载: 导出CSV
表 2 流型图像数据集
Table 2. Flow pattern images dataset
编号 滑油流量/
（kg/h）空气流量/
（L/min）流型图像 流型 1 179.01 1.82 
弹状流 2 178.73 4.65 
弹状流 $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ 240 175.73 11.78 
弹状流 241 225.04 48.5 
层状流 242 286.74 65.4 
层状流 $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ 480 354.61 75.4 
层状流 481 19.215 49.4 
波状流 482 18.55 77.5 
波状流 $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ 720 18.59 119 
波状流 721 141.65 374 
环状流 722 177.50 240 
环状流 $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ 960 176.45 377 
环状流
下载: 导出CSV
表 3 流型特征向量数据集
Table 3. Flow pattern eigenvector dataset
数据集 序号 属性 标记 λ1 λ2 λ3 λ4 M1 S1 H 训练集 1 0.27602 0.05488 0.54859 0.24524 0.36672 0.29988 0.12573 [1 0 0 0] 2 0.31042 0.06080 0.56414 0.23159 0.39088 0.45241 0.09634 [1 0 0 0] $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ 168 0.07642 0.01122 0.21356 0.08093 0.07086 0.34329 0.24814 [1 0 0 0] 169 0.51202 0.51534 0.37231 0.64693 0.69113 0.22481 0.20317 [0 1 0 0] 170 0.45004 0.49476 0.32314 0.62141 0.68798 0.22291 0.20814 [0 1 0 0] $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ 336 0.59299 0.74805 0.40717 0.79587 0.58264 0.22434 0.23388 [0 1 0 0] 337 0.56652 0.75315 0.36767 0.88096 0.64067 0.21653 0.33021 [0 0 1 0] 338 0.61191 0.95224 0.38447 0.95254 0.57249 0.30584 0.32546 [0 0 1 0] $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ 504 0.60362 0.99469 0.39109 1.00000 0.54537 0.32531 0.34671 [0 0 1 0] 505 0.46931 0.46303 0.36499 0.70119 0.63918 0.45775 0.40734 [0 0 0 1] 506 0.53189 0.57443 0.38286 0.75554 0.59591 0.33784 0.41739 [0 0 0 1] $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ 672 0.43471 0.44928 0.33359 0.69478 0.65760 0.43377 0.37798 [0 0 0 1] 测试集 673 0.27133 0.09487 0.32741 0.12510 0.41909 0.95555 0.10750 [1 0 0 0] $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ 960 0.41931 0.52469 0.30799 0.71374 0.16645 0.22922 0.71595 [0 0 0 1]
下载: 导出CSV
表 4 部分样本辨识结果
Table 4. Partial sample identification results
序号 属性 真实标签 分类结果 λ1 λ2 λ3 λ4 M1 S1 H 673 0.35485 0.24958 0.25443 0.34721 0.45807 0.12344 0.21527 [1 0 0 0] [1 0 0 0] 674 0.28727 0.30627 0.33927 0.50878 0.24334 0.63038 0.26157 [1 0 0 0] [1 0 0 0] $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ 744 0.29919 0.06911 0.43866 0.11674 0.41688 0.95434 0.11284 [1 0 0 0] [1 0 0 0] 745 0.53133 0.54798 0.48986 0.69045 0.26387 0.31063 0.18983 [0 1 0 0] [0 1 0 0] 746 0.48897 0.59400 0.42638 0.70242 0.32896 0.31190 0.15950 [0 1 0 0] [0 1 0 0] $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ 816 0.47195 0.54364 0.43173 0.67097 0.35959 0.29581 0.16097 [0 1 0 0] [0 1 0 0] 817 0.34700 0.17236 0.50852 0.40125 0.86627 0.43184 0.10776 [0 0 1 0] [0 0 1 0] 818 0.36476 0.15776 0.53024 0.40888 0.92262 0.43356 0.10559 [0 0 1 0] [0 0 1 0] $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ 888 0.35594 0.20106 0.48196 0.45726 0.93192 0.55810 0.11376 [0 0 1 0] [0 0 1 0] 889 0.37314 0.47692 0.28547 0.71207 0.13399 0.31519 0.86148 [0 0 0 1] [0 0 0 1] 890 0.44483 0.48485 0.34966 0.71330 0.25744 0.25888 0.59261 [0 0 0 1] [0 0 0 1] $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ $ \vdots $ 960 0.40420 0.51042 0.34401 0.74687 0.29240 0.17179 0.55442 [0 0 0 1] [0 0 0 1]
下载: 导出CSV
-
[1] 刘振侠,江平. 航空发动机机械系统设计[M]. 北京: 科学出版社,2022. LIU Zhenxia,JIANG Ping. Mechanical system design of aero-engine[M]. Beijing: Science Press,2022. (in ChineseLIU Zhenxia, JIANG Ping. Mechanical system design of aero-engine[M]. Beijing: Science Press, 2022. (in Chinese) [2] FLOUROS M,KANARACHOS A,YAKINTHOS K,et al. Two-phase flow pressure drop in corrugated tubes used in an aero-engine oil system[J]. Journal of Engineering for Gas Turbines and Power,2016,138(6): 062603. doi: 10.1115/1.4031627 [3] KANARACHOS S,FLOUROS M. Simulation of the air-oil mixture flow in the scavenge pipe of an aero engine using generalized interphase momentum exchange models[J]. WSEAS Transactions on Fluid Mechanics,2014,9: 144-153. [4] 陈家琅,陈涛平. 石油气液两相管流[M]. 2版. 北京: 石油工业出版社,2010. CHEN Jialang,CHEN Taoping. Liquid-liquid two-phase pipeline flow of petroleum gas[M]. 2nd ed. Beijing: Petroleum Industry Press,2010. (in ChineseCHEN Jialang, CHEN Taoping. Liquid-liquid two-phase pipeline flow of petroleum gas[M]. 2nd ed. Beijing: Petroleum Industry Press, 2010. (in Chinese) [5] MANDHANE J M,GREGORY G A,AZIZ K. A flow pattern map for gas-liquid flow in horizontal pipes[J]. International Journal of Multiphase Flow,1974,1(4): 537-553. doi: 10.1016/0301-9322(74)90006-8 [6] WEISMAN J,DUNCAN D,GIBSON J,et al. Effects of fluid properties and pipe diameter on two-phase flow patterns in horizontal lines[J]. International Journal of Multiphase Flow,1979,5(6): 437-462. doi: 10.1016/0301-9322(79)90031-4 [7] SPEDDING P L,NGUYEN V T. Regime maps for air water two phase flow[J]. Chemical Engineering Science,1980,35(4): 779-793. doi: 10.1016/0009-2509(80)85062-7 [8] TAITEL Y,DUKLER A E. A model for predicting flow regime transitions in horizontal and near horizontal gas-liquid flow[J]. AIChE Journal,1976,22(1): 47-55. doi: 10.1002/aic.690220105 [9] BARNEA D,SHOHAM O,TAITEL Y. Flow pattern transition for downward inclined two phase flow; horizontal to vertical[J]. Chemical Engineering Science,1982,37(5): 735-740. doi: 10.1016/0009-2509(82)85033-1 [10] HUBBARD M G. An analysis of horizontal gas-liquid slug flow[D]. Houston,US: University of Houston,1965. [11] 周云龙,孙斌,李岩,等. 倾斜下降管内气-液两相流流型PSD特征[J]. 热科学与技术,2004,3(2): 129-132. ZHOU Yunlong,SUN Bin,LI Yan,et al. PSD characteristics of gas-liquid two-phase flow pattern in inclined downward tube[J]. Journal of Thermal Science and Technology,2004,3(2): 129-132. (in ChineseZHOU Yunlong, SUN Bin, LI Yan, et al. PSD characteristics of gas-liquid two-phase flow pattern in inclined downward tube[J]. Journal of Thermal Science and Technology, 2004, 3(2): 129-132. (in Chinese) [12] 纪晨晨. 电容层析成像的2D图像重建及流型识别算法研究[D]. 西安: 西安科技大学,2021. JI Chenchen. Research on 2D image reconstruction and flow pattern recognition of electrical capacitance tomography[D]. Xi’an: Xi’an University of Science and Technology,2021. (in ChineseJI Chenchen. Research on 2D image reconstruction and flow pattern recognition of electrical capacitance tomography[D]. Xi’an: Xi’an University of Science and Technology, 2021. (in Chinese) [13] 方立德,张垚,张万岭,等. 基于声发射技术的垂直管气液两相流动检测方法[J]. 化工学报,2014,65(4): 1243-1250. FANG Lide,ZHANG Yao,ZHANG Wanling,et al. Flow detection technology based on acoustic emission of gas-liquid two-phase flow in vertical pipe[J]. CIESC Journal,2014,65(4): 1243-1250. (in ChineseFANG Lide, ZHANG Yao, ZHANG Wanling, et al. Flow detection technology based on acoustic emission of gas-liquid two-phase flow in vertical pipe[J]. CIESC Journal, 2014, 65(4): 1243-1250. (in Chinese) [14] BUI DINH T,CHOI T S. Application of image processing techniques in air/water two phase flow[J]. Mechanics Research Communications,1999,26(4): 463-468. doi: 10.1016/S0093-6413(99)00050-6 [15] 施丽莲. 基于数字图像识别技术的气液两相流参数检测的研究[D]. 杭州: 浙江大学,2004. SHI Lilian. Study on parameter measurement of gas-liquid two-phase flow based on digital image recognition[D]. Hangzhou: Zhejiang University,2004. (in ChineseSHI Lilian. Study on parameter measurement of gas-liquid two-phase flow based on digital image recognition[D]. Hangzhou: Zhejiang University, 2004. (in Chinese) [16] WANG Hongyi,DONG Feng. Image features extraction of gas/liquid two-phase flow in horizontal pipeline by GLCM and GLGCM[C]//Proceedings of the 9th International Conference on Electronic Measurement and Instruments. Piscataway,US: IEEE,2009: 2.135-2.139. [17] ZHANG Wenyin,SHIH F Y,JIN Ningde,et al. Recognition of gas-liquid two-phase flow patterns based on improved local binary pattern operator[J]. International Journal of Multiphase Flow,2010,36(10): 793-797. doi: 10.1016/j.ijmultiphaseflow.2010.06.002 [18] SHANTHI C,PAPPA N,ASWINI SUGANYA J. Digital image processing based flow regime identification of gas/liquid two-phase flow[J]. IFAC Proceedings Volumes,2013,46(32): 409-414. doi: 10.3182/20131218-3-IN-2045.00170 [19] ZHOU Yunlong,CHEN Fei,SUN Bin. Identification method of gas-liquid two-phase flow regime based on image multi-feature fusion and support vector machine[J]. Chinese Journal of Chemical Engineering,2008,16(6): 832-840. doi: 10.1016/S1004-9541(09)60002-1 [20] 仝卫国,庞雪纯,朱赓宏. 基于卷积神经网络的气液两相流流型识别方法[J]. 系统仿真学报,2021,33(4): 883-891. TONG Weiguo,PANG Xuechun,ZHU Genghong. Gas-liquid two-phase flow pattern recognition method based on convolutional neural network[J]. Journal of System Simulation,2021,33(4): 883-891. (in ChineseTONG Weiguo, PANG Xuechun, ZHU Genghong. Gas-liquid two-phase flow pattern recognition method based on convolutional neural network[J]. Journal of System Simulation, 2021, 33(4): 883-891. (in Chinese) [21] 刘俊宏. 基于图像处理和小波卷积神经网络的气液两相流流型识别方法[D]. 吉林: 东北电力大学,2022. LIU Junhong. Flow pattern recognition method for gas-liquid two-phase flow based on image processing and wavelet convolution neural network[D]. Jilin: Northeast Dianli University,2022. (in ChineseLIU Junhong. Flow pattern recognition method for gas-liquid two-phase flow based on image processing and wavelet convolution neural network[D]. Jilin: Northeast Dianli University, 2022. (in Chinese) [22] 陈飞. 基于数字图像处理的气液两相流流型智能识别方法[D]. 吉林: 东北电力大学,2008. CHEN Fei. The identification method of gas-liquid two-phase flow regime based on digital image processing[D]. Jilin: Northeast Dianli University,2008. (in ChineseCHEN Fei. The identification method of gas-liquid two-phase flow regime based on digital image processing[D]. Jilin: Northeast Dianli University, 2008. (in Chinese) [23] 李蕴奇. 基于小波变换的图像阈值去噪及其效果评估[J]. 东北师大学报(自然科学版),2012,44(1): 60-66. LI Yunqi. Image threshold de-noising based on wavelet transform and performance evaluation[J]. Journal of Northeast Normal University (Natural Science Edition),2012,44(1): 60-66. (in ChineseLI Yunqi. Image threshold de-noising based on wavelet transform and performance evaluation[J]. Journal of Northeast Normal University (Natural Science Edition), 2012, 44(1): 60-66. (in Chinese) [24] MALLAT S G. A theory for multiresolution signal decomposition: the wavelet representation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,1989,11(7): 674-693. doi: 10.1109/34.192463 [25] 柳小桐. BP神经网络输入层数据归一化研究[J]. 机械工程与自动化,2010(3): 122-123,126. LIU Xiaotong. Study on data normalization in BP neural network[J]. Mechanical Engineering & Automation,2010(3): 122-123,126. (in ChineseLIU Xiaotong. Study on data normalization in BP neural network[J]. Mechanical Engineering & Automation, 2010(3): 122-123, 126. (in Chinese) [26] ELMAN J. Finding structure in time[J]. Cognitive Science,1990,14(2): 179-211. doi: 10.1207/s15516709cog1402_1 [27] 丛爽,高雪鹏. 几种递归神经网络及其在系统辨识中的应用[J]. 系统工程与电子技术,2003,25(2): 194-197. CONG Shuang,GAO Xuepeng. Recurrent neural networks and their application in system identification[J]. Systems Engineering and Electronics,2003,25(2): 194-197. (in ChineseCONG Shuang, GAO Xuepeng. Recurrent neural networks and their application in system identification[J]. Systems Engineering and Electronics, 2003, 25(2): 194-197. (in Chinese) [28] 丁硕,常晓恒,巫庆辉,等. Elman和BP神经网络在模式分类领域内的对比研究[J]. 现代电子技术,2014,37(8): 12-14,18. DING Shuo,CHANG Xiaoheng,WU Qinghui,et al. Comparative study of Elman and BP neural networks used for pattern classification[J]. Modern Electronics Technique,2014,37(8): 12-14,18. (in ChineseDING Shuo, CHANG Xiaoheng, WU Qinghui, et al. Comparative study of Elman and BP neural networks used for pattern classification[J]. Modern Electronics Technique, 2014, 37(8): 12-14, 18. (in Chinese) [29] 张开放,苏华友,窦勇. 一种基于混淆矩阵的多分类任务准确率评估新方法[J]. 计算机工程与科学,2021,43(11): 1910-1919. ZHANG Kaifang,SU Huayou,DOU Yong. A new multi-classification task accuracy evaluation method based on confusion matrix[J]. Computer Engineering and Science,2021,43(11): 1910-1919. (in ChineseZHANG Kaifang, SU Huayou, DOU Yong. A new multi-classification task accuracy evaluation method based on confusion matrix[J]. Computer Engineering and Science, 2021, 43(11): 1910-1919. (in Chinese) [30] 周志华. 机器学习[M]. 北京: 清华大学出版社,2016. ZHOU Zhihua. Machine learning[M]. Beijing: Tsinghua University Press,2016. (in ChineseZHOU Zhihua. Machine learning[M]. Beijing: Tsinghua University Press, 2016. (in Chinese) -
点击查看大图
图(10) / 表(4)
计量
- 文章访问数: 978
- HTML浏览量: 291
- PDF量: 56
- 被引次数: 0