Flow pattern recognition method of gas-liquid two-phase flow based on adaptive optimal kernel and convolution neural network

doi:10.11949/j.issn.0438-1157.20180198

Abstract

Abstract:

To study the dynamic characteristics of gas-liquid two-phase flow, and to solve the traditional problem that the eigenvalues extracted are so few that is not representative and result in low recognition rate, V-cone flow-meter and dynamic differential pressure sensor were used to obtain fluctuating signals of gas-liquid two-phase flow in different flow patterns, time-frequency analysis of obtained dynamic signals were performed by adaptive optimal kernel(AOK) algorithm, to convert the one-dimensional time-domain signals to three-dimensional time-frequency spectra so as to describe clearly the flow status of gas-liquid two-phase flow in the pipeline. Time-frequency spectra of different flow patterns were learned by the convolution neural network (CNN) to extract the corresponding eigenvalue automatically, then were practiced by Softmax classifier to achieve flow pattern recognition. By experiment and analysis for several common flow patterns, adopting time-frequency spectrum and deep learning method of convolution neural network was found to identify the gas-liquid two-phase flow pattern can overcome the shortcomings of the traditional flow pattern recognition method for few eigenvalue extracted, and describe appropriately the dynamic characteristics of gas-liquid two-phase flow. This method allows further investigation of more types of flow patterns as well as porosity and more.

Key words: gas-liquid flow, flow pattern recognition, algorithm, time-frequency analysis, neural networks

摘要：

为了研究气液两相流的动态特性，以及解决提取的特征值少而没有代表性导致识别率不高的传统问题，利用V锥流量计和动态差压传感器获取气液两相流在不同流型下的波动信号，采用自适应最优核算法对获取的动态信号进行时频分析，把一维时域信号转换为三维的时频谱图，能够清晰描述出管道内气液两相流的流动状态。将不同流型的时频谱图通过卷积神经网络（CNN）进行学习并自动提取相应的特征值，然后使用Softmax分类器进行训练从而实现流型识别。通过对几种常见流型进行试验与分析发现，采用时频谱图结合卷积神经网络的深度学习方法识别气液两相流流型，克服了传统流型识别方法特征值提取的不足之处，能够更贴切地描述气液两相流的动态特征。此方法可以进一步研究更多种类的流型以及空隙率等。

关键词: 气液两相流, 流型识别, 算法, 时频分析, 神经网络

CLC Number:

TP274.2

WENG Runying, SUN Bin, ZHAO Yuxiao, ZHANG Jingyue, WEN Yingjie. Flow pattern recognition method of gas-liquid two-phase flow based on adaptive optimal kernel and convolution neural network[J]. CIESC Journal, 2018, 69(12): 5065-5072.

翁润滢, 孙斌, 赵玉晓, 张竟月, 文英杰. 基于自适应最优核和卷积神经网络的气液两相流流型识别方法[J]. 化工学报, 2018, 69(12): 5065-5072.

References 30

[1]	杨秋怡, 韩云峰, 金宁德. 聚合物作用下油水两相流油泡时频运动特性[J]. 工程热物理学报, 2016, 37(27):2582-2589. YANG Q Y, Han Y F, JIN N D. Time-frequency characteristic of oil droplets in oil-water two-phase flow with polymer addition[J]. Journal of Engineering Thermophysis, 2016, 37(37):2582-2589.
[2]	LAN J, SHEN L H. Wavelet analysis of pressure fluctuation in a bubbling fluidized bed[J]. Boiler Technology, 2003, 34(4):42-46.
[3]	GAO Z, JIN N. Flow-pattern identification and nonlinear dynamics of gas-liquid two-phase flow in complex networks[J]. Phys. Rev. E Stat. Nonlin. Soft Matter Phys., 2009, 79(6):066303.
[4]	周云龙, 顾杨杨. 基于独立分量分析和RBF神经网络的气液两相流流型识别[J]. 化工学报, 2012, 63(3):796-799. ZHOU Y L, GU Y Y. Flow regime identification of gas-liquid twophase flow based ICA and RBF neural networks[J]. CIESC Journal, 2012, 63(3):796-799.
[5]	郑桂波, 金宁德. 两相流流型多尺度熵及动力学特性分析[J]. 物理学报, 2009, 58(7):4485-4492. ZHENG G B, JIN N D. Multiscale entropy and dynamic characteristics of two-phase flow patterns[J]. Physical Review Letters, 2009, 58(7):4485-4492.
[6]	WU H J, ZHOU F D, WU Y Y. Intelligent identification system of flow regime of oil/gas/water[J]. Int. J. Multiphase Flow, 2001, 27:459-475.
[7]	周云龙, 刘旭, 李洪伟. 2种熵测度在分析流型信号复杂度上的应用[J].化学工程, 2011, 39(8):53-56. ZHOU Y L, LIU X, LI H W. Application of two entropy measures to analyzing complexity of multiphase flow signal[J]. Chemical Engineering(China), 2011, 39(8):53-56.
[8]	周云龙, 尹洪梅, 丁会晓. 多尺度熵在棒束通道气液两相流压差信号分析中的应用[J]. 化工学报, 2016, 67(9):3625-3631. ZHOU Y L, YIN H M, DING H X. Application of multi-scale entropy in analyzing pressure difference signals of gas-liquid two-phase flow in rod bundled channel[J]. CIESC Journal, 2016, 67(9):3625-3631.
[9]	李文升, 郭烈锦, 谢晨. 集输立管内气水两相流压差信号的特征分析[J]. 工程热物理学报, 2015, 36(6):1247-1251. LI W S, GUO L J, XIE C. Analysis on the signal features of differential pressure for air-water two-phase flow in pipeline-riser system[J]. Journal of Engineering Thermophysics, 2015, 36(6):1247-1251.
[10]	COSTA M, GOLDBERGER A L, PENG C K. Multiscale entropy analysis of complex physiologic time series[J]. Physical Review Letters, 2002, 89(6):068102.
[11]	RICHMAN J S, MOORMAN J R. Physiological time-series analysis using approximate entropy and sample entropy[J]. American Journal of Physiology-Heart and Circulatory Physiology, 2000, 278(6):H2039-H2049.
[12]	李洪伟, 周云龙, 宋倩, 等. 水平气液两相流流型图像信息递归特征分析[J]. 化工学报, 2010, 61(6):1431-1436. LI H W, ZHOU Y L, SONG Q, et al. Recursive feature of gas-liquid two-phase flow pattern based on information entropy series of flow image[J]. CIESC Journal, 2010, 61(6):1431-1436.
[13]	杨靖, 郭烈锦. 气液两相流压差信号的非线性分析[J]. 中国电机工程学报, 2002, 22(7):134-139. YANG J, GUO L J. A nonlinear analysis on differential pressure drop signal for gas-liquid two-phase flow[J]. Proceedings of the CSEE, 2002, 22(7):134-139.
[14]	段艳杰, 吕宜生, 张杰, 等. 深度学习在控制领域的研究现状与展望[J]. 自动化学报, 2016, 42(5):643-654. DUN Y J, LU Y S, ZHANG J, et al. Deep learning for control:the state of the art and prospects[J]. Acta Automatica Sinica, 2016, 42(5):643-654.
[15]	SCHMIDHUBER J. Deep learning in neural networks:an overview[J]. Neural Networks, 2015, 61:85-117.
[16]	HINTON G E, OSINDERO S, TEH Y W. A fast learning algorithm for deep belief nets[J]. Neural Computation, 2006, 18(7):1527-1554.
[17]	王伟凝, 王励, 赵明权, 等. 基于并行深度卷积神经网络的图像美感分类[J]. 自动化学报, 2016, 42(6):904-914. WANG W N, WANG L, ZHAO M Q, et al. Image aesthestic classification using parallel deep convolutional neural networks[J]. Acta Automatica Sinica, 2016, 42(6):904-914.
[18]	MEI S, JI J, HOU J, et al. Learning sensor-specific spatial-spectral features of hyperspectral images via convolutional neural networks[J]. IEEE Transactions on Geoscience & Remote Sensing, 2017, 55(8):4520-4533.
[19]	张帅. 基于深度学习的植物叶片识别算法研究[D]. 北京:北京林业大学, 2016. ZHANG S. Research on plant leaf images identification algorithm based on deep learning[D]. Beijing:Beijing Forestry University, 2016.
[20]	AREL I, ROSE D C, KARNOWSKI T P. Deep machine learning a new frontier in artificial intelligence research[J]. IEEE Computational Intelligence Magazine, 2010, 5(4):13-18.
[21]	HINTON G E, SALAKHUTDINOV R R. Reducing the dimensionality of data with neural networks[J]. Science, 2006, 313(5786):504-507.
[22]	孙斌, 王二朋, 郑永军. 气液两相流波动信号的时频谱分析研究[J]. 物理学报, 2011, 60(1):381-388. SUN B, WANG E P, ZHENG Y J. Time-frequency spectral analysis of gas-liquid two-phase flow's fluctuation[J]. Acta Physica Sinica, 2011, 60(1):381-388.
[23]	孙斌, 张宏建, 岳伟挺. HHT与神经网络在油气两相流流型识别中的应用[J]. 化工学报, 2004, 55(10):1723-1727. SUN B, ZHANG H J, YUE W T. Applied study of HHT and neural networks on flow regime identification for oil-gas two-phase flow[J]. Journal of Chemical Engineering and Industry (China), 2004, 55(10):1723-1727.
[24]	王秀芳, 汪清, 李博健. 基于自适应最优核时频分析的管道泄漏检测应用研究[J]. 化工自动化及仪表, 2015, 42(7):770-773+849. WANG X F, WANG Q, LI B J. Research of gas pipeline leak detection based on adaptive optimal kernel time-frequency analysis[J]. Control and Instruments in Chemical Industry, 2015, 42(7):770-773+849
[25]	王晓凯, 高静怀, 何洋洋. 基于时频自适应最优核的时频分析方法[J]. 系统工程与电子技术, 2010, 32(1):22-26. WANG X K, GAO J H, HE Y Y. Time-frequency analysis based on time-frequency-adaptive optional-kernel[J]. Systems Engineering and Electronics, 2010, 32(1):22-26.
[26]	COIFMAN R R, DONOHO D L. Translation-Invariant De-Noising[M]. New York:Springer, 1995:125-150.
[27]	LECUN Y, BOTTOU L, BENGIO Y, et al. Gradient-based learning applied to document recognition[J]. Proceedings of the IEEE, 1998, 86(11):2278-2324.
[28]	HOPFIELD J J. Neural networks and physical systems with emergent collective computational abilities[J]. Proceedings of the National Academy of Sciences of the United States of America, 1982, 79(8):2554-2558.
[29]	SUN H Q, PANG Y W. Glance nets-efficient convolutional neural networks with adaptive hard example mining[J]. Science China(Information Sciences), 2018, 61(10):253-255.
[30]	BENGIO Y. Learning deep architectures for AI[J]. Foundations and Trends in Machine Learning, 2009, 2(1):1-12.