پیش‌بینی میزان آلایندة NOx در یک محفظة احتراق آزمایشگاهی با شعلة پایدارشدة چرخشی با استفاده از روش پردازش تصویر شعله و به‌کارگیری روش‌های داده‌کاوی

نوع مقاله: مقاله پژوهشی

نویسندگان

1 عضو هیات علمی / دانشکدة علوم و فنون نوین، دانشگاه تهران

2 دانشجوی کارشناس ارشد / دانشکدة علوم و فنون نوین، دانشگاه تهران

چکیده

هدف این مقاله پیش‌بینی محدودة آلایندة NOx براساس ویژگی‌های استخراج‌شده از تصویر شعله (شامل ویژگی‌های هندسی و نورتابی) است. در این مقاله ارتباط بین تصاویر ثبت‌شده از شعله در نقاط مختلف عملکردی یک محفظة احتراق آزمایشگاهی با کاربرد در توربین‌های گازی نیروگاهی، با مقادیر اندازه‌گیری‌شدة سطح آلایندة NOx تولیدشده در این نقاط عملکردی، به‌کمک پردازش تصویر شعله و به‌کارگیری سه روش شبکة عصبی متفاوت بررسی شده است. ویژگی‌های استخراج‌شده از تصویر شعله به‌عنوان ورودی به شبکه‌های عصبی LVQ، خودسازمانده و چندلایه ارائه و بر این اساس محدودة مقادیر NOx مربوط به تصاویر شعله، پیش‌بینی و با مقادیر اندازه‌گیری‌شده از محفظه صحت‌سنجی می‌شود. آزمایشات مربوطه با به‌کارگیری چهار نوع انژکتور پاشش سوخت ثانویه، با ساختار هندسی و طراحی متفاوت، در شرایط نسبت هم‌ارزی کلی در محدودة 0/7 تا 0/9 همراه با مقادیر مختلف دبی‌ پاشش سوخت ثانویه در محدودة صفر تا 4/2 لیتر بر دقیقه قبلاً انجام شده است. نتایج نشان می‌دهد که شبکة عصبی LVQ جهت پیش‌بینی میزان آلایندگی NOx با دقت 97 درصد توانایی بالاتری نسبت به شبکة عصبی چندلایه (با دقت 95 درصد) و خودسازمانده (با دقت 89 درصد) دارد. نوآوری این پژوهش در این است که تاکنون روی این محفظة احتراق (همراه با پاشش سوخت ثانویه) هیچ مطالعه‌ای بر پایة پردازش تصویر شعله انجام نشده است.

کلیدواژه‌ها

موضوعات


عنوان مقاله [English]

Prediction of NOx emission of an experimental swirl stabilized combustor using the flame image processing techniques and data mining methods

نویسندگان [English]

  • Alireza Torabi 1
  • Rouzbeh Riazi 1
  • Mohamad Daneshi Kohani 2
  • Shidvash Vakilipour 1
  • Hadi Veisi 1
  • Hadi Zare 1
چکیده [English]

So far, some methods have been presented for prediction of parameters of a combustion system based on the flame image processing techniques. Almost all of them use various techniques to convert the extracted geometrical and luminosity data of flame image into combustion information. Potentially, image of a flame includes many information in terms of different features which could be related to the combustion field. In this work, relations between recorded images taken from turbulent flames in a combustion chamber with measured values of NOx emission have been investigated using image processing techniques along with three methods of data mining. For this purpose geometrical, statistical and luminosity features extracted from flame images were used as input data for the LVQ, MLP and SOM neural networks. Based on these extracted features from flame images the neural networks predicted the level of NOx emission and these predicted values were validated with the measured data of combustor. Moreover by using forward feature selection technique in each of the above-mentioned algorithms, five features were selected. The related experiments were already performed by using four different types of secondary fuel injectors (with four different designs) for an overall equivalence ratio between =0.7~0.9 along with different amount of secondary fuel injection rate in the range of Qsec=0.6~4.2 L/min. The result shows that the LVQ with 97% accuracy has better capability for prediction of the level of NOx emission than SOM and MLP methods.

کلیدواژه‌ها [English]

  • Swirl-stabilized combustor
  • MLP
  • SOM neural network
  • LVQ neural network

[1] R. Hernández, J. Ballester, Flame imaging as a diagnostic tool for industrialn combustion, Journal of Combustion and Flame, Vol. 15, pp. 509-528, 2008.

[2] F. Jiang, S. Liu, S. Liang, Z. Li, X. Wang, G. Lu, Visual flame monitoring system based on two-color method, Journal of Thermal Science, Vol 18, pp. 284-288, 2009.

[3] J. Ballester, T. Garca-Armingol, Diagnostic techniques for the monitoring and control of practical flames, Prog Energy Combust, Vol. 36, pp. 375-411, 2010.

[4] G. Lu, Y. Yan, Y. Huang, A. Reed, An Intelligent Vision System for Monitoring and Control of Combustion Flames, Journal of Measurement and Control, Vol. 32, No. 6, pp. 164-168, 1999.

[5] J. Wua, Z. Ming-chuan, F. Hao-Jie, F. Wei-dong, Z. Yue-gui, A study on fractal characteristics of aerodynamic field in low-NOx coaxial swirling burner, Journal of Chemical Engineering Science, Vol. 59, No. 7, pp. 1473-1479, 2004.

[6] A. González-Cencerrado, B. Peña, A. Gil, Coal flame characterization by means of digital image processing in a semi-industrial scale PF swirl burner, Applied Energy, Vol. 94, pp. 375-384, 2012.

[7] A. González-Cencerrado, B. Peña, Characterization of PF flames under different swirl conditions based on visualization systems, Journal of Fuel, Vol. 113, pp. 798-809, 2013.

[8] A. González-Cencerrado, B. Peña, A. Gil, Experimental analysis of biomass co-firingflames in a pulverized fuel swirl burner using a CCD based visualization system, Journal of Fuel Processing Technology, Vol. 130, pp. 299-310, 2015.

[9] M. G. A. Rahman, J. R. Gibbins, A. K. Forrest, Combustion in power station boilers-Advanced monitoring using imaging, Project report, Cleaner Fossil Fuels Programme, 2004.

[10] F. Tang, Z. Kongjin, D. Mansheng, S. Qing, Mean flame height and radiative heat flux characteristic of medium scale rectangular thermal buoyancy source with different aspect ratios in a sub-atmospheric pressure, International Journal of Heat and Mass Transfer, Vol. 84, pp. 427-432, 2015.

[11] H. Yu, J. F. MacGregor, Monitoring flames in an industrial boiler using multivariate image analysis, American Institute of Chemical Engineers, Vol. 50, No. 7, pp. 1474-1483, 2004.

[12] A. Tuntrakoon, S. Kuntanapreeda, Image-based flame control of a premixed gas burner using fuzzy logics, foundations of intelligent systems, springer-verlag, Berlin/Heidelberg, pp. 673-677, 2003.

[13] F. Wang, X. J. Wang, Z. Y. Ma, J. H. Yan, Y. Chi, C. Y. Wei, M. J. Ni, K. F. Cen, The research on the estimation for the NOx emissive concentration of the pulverized coal boiler by the flame image processing technique, Journal of Fuel, Vol. 81, pp. 2113-2120, 2003.

[14] W. Yan, C. Lu, M. Colechin, Monitoring and characterisation of pulverised coal flames using digital imaging techniques, Journal of Fuel, Vol. 81, No. 5, pp. 647–656, 2002.

[15] W. Yan, L. Chun, Two-dimensional distributions of temperature & soot volume fraction inversed from visible flame images, Experimental Thermal and Fluid Science, Vol. 50, pp. 229-233, 2013.

[16] D. Sbarbaro, O. Farias, Z. Zawadsky, Real-time monitoring and characterization of flames by principal-component analysis, Journal of Combustion and Flame, Vol. 132, pp. 591-595, 2003.

[17] H. W. Huang, Z. Yang, Imaging based chemiluminescence characterisation of partially premixed syngas flames through DFCD technique, International Journal of Hydrogen Energy, Vol. 38, pp. 4839-4847. 2013.

[18] J. Chen, L. Chan, Y. C. Cheng, Gaussian process regression based optimal design of combustion systems using flame images, Applied Energy, Vol. 111, pp. 153-160, 2013.

[19] B. Lin, S. B. Jorgensen, Soft sensor design by multivariate fusion of image features and process measurements, Journal of Process Control, Vol. 21, pp. 347-553, 2011.

[20] J. S. Wang, X. D. Ren, GLCM based extraction of flame image texture features and KPCA-GLVQ recognition method for rotary kiln combustion working conditions, International Journal of Automation and Computing, Vol. 11, No. 1, pp. 72-77, 2014.

[21] H. Zhou, Q. Tang, L. Yang, Y. Yan, G. Lu, K. Cen, Support vector machine based online coal identification through advanced flame monitoring, Journal of Fuel, Vol. 117, pp. 944-951, 2014.

[22] R. Riazi, M. Farshchi, M. Shimura, M. Tanahashi, T. Miyauchi, An Experimental Study on Combustion Dynamics and NOx Emission of a Swirl Stabilized Combustor with Secondary Fuel Injection, Journal of Thermal Science and Technology, Vol. 5, No. 2, pp. 266-281, 2010.

[23] M. Shimura, M. Tanahashi, G. Choi, T. Miyauchi, Large-scale vortical motion and pressure fluctuation in noise-controled swirl-stabilized combustor, Journal of Thermal Science and Technology, Vol. 4, No. 4, pp. 494-506, 2009.

[24] M. Tanahashi, S. Inoue, M. Shimura, S. Taka, G. Choi, T. Miyauchi, Reconstructed 3D flame structures in noise-controlled swirl-stabilized combustor, Experiments in Fluids, Vol. 45, No. 3, pp. 447-460, 2008.

[25] M. Tanahashi, S. Murakami, T. Miyauchi, G. Choi, Control of oscillating combustion and measurements of turbulent flames, Proceedings of 5th Symposium on Smart Control of Turbulence, pp. 75-84, 2004.

[26] G. Choi, M. Tanahashi, T. Miyauchi, Control of oscillating combustion and noise based on local flame structure, Proceedings of the Combustion Institute, Vol. 30, No. 2, pp. 1807-1814, 2005.

[27] E. R. Davis, Machine Vision, University of Southern Queensland, pp. 20-22, 1990.

[28] D. N. Joanes, C. A. Gill, Comparing measures of sample skewness and kurtosis, Journal of the Royal Statistical Society (Series D), Vol. 47, No. 1, pp 183-189, 1998.

[29] Y. Cao, J. Wu, M. I. Jianchun, Z. Yu, Flame structure of a jet flame with penetration of side micro-jets, Chinese Journal of Chemical Engineering, Vol. 16, No. 6, pp. 861-866, 2008.

[30] M. Hagan, H. Demuth, M. Beale, Neural Network Design, Boston: PWS Publishing, 1996.

[31] M. T. Vakil-Baghmisheh, N. Pavešic, Premature clustering phenomenon and new training algorithms for LVQ, Journal of Pattern Recognition, Vol. 36, pp. 1901-1921, 2003.

[32] J. Keller, I. Hongo, Pulse Combustion: The mechanisms of NOx production, Combustion and Flame, Vol. 80, pp. 219-237, 1990.

[33] A. H. Lefebvre, Fuel effects on gas turbine combustion, Final Report, Air Force Wright Aeronautical Laboratories, 2004.

[34] J. Odgers, D. Kretschmer, The prediction of thermal NOx in gas turbines, ASME Paper 85-1GT-126, 1985.

[35] S. R. Turns, Introduction to combustion, 2nd ed., New York: McGraw-Hill, 2000.

[36] M. LaViolette, R. Perez, On the prediction of pollutant emission indices from gas turbine combustion chambers, Proceedings of ASME Turbo Expo, 2012.

[37] T. S. Cheng, Effects of partial premixing on pollutant emissions in swirling methane jet flames, Journal of Combustion and Flame, Vol. 125, No. 1-2, pp. 865-878, 2001.

[38] G. A. Lavoie, A. F. Schlader, A xsaling study of NO formation in turbulent flames of hydrocarbon burning in air, Journal of Combustion Science and Technology, Vol. 8, pp. 215-224, 1973.

[39] N. Peters, S. Donnerhack, 18th Symposium (International) on Combustion, the Combustion Institute, Pittsburgh, pp. 33, 1981.