差压式重介悬浮液密度与粘度一体化测量方法研究
|
摘要
密度、粘度是重介质选煤过程中需要准确测量的两项重要工艺参数。目前,一般采用同位素密度计检测重介悬浮液的密度,但同位素密度计的辐射性对人体有害。此外,工业现场没有重介悬浮液粘度的在线测量设备,而是先用磁性物含量计测量悬浮液中的磁铁粉含量,再结合同位素密度计测量到的密度值,计算出煤泥含量,用煤泥含量作为评价粘度的指标。然而煤泥含量在线测量误差较大,不少工厂仍采用离线化验的方法。
本文提出一种差压式重介悬浮液密度与粘度一体化测量方案,实现了密度无核测量以及粘度的在线检测。该方案在悬浮液垂直主回路附近开辟一旁路,在旁路上安装一套差压传感器,当悬浮液在旁路中稳定流动时,通过差压输出测量密度;关闭旁路上、下阀门,重介悬浮液在旁路中发生静态沉降,通过沉降过程测量其粘度。
本文的目的就是通过实验对测量方案进行验证,并给出从悬浮液沉降曲线到粘度的建模方法。实验分为两部分,第一部分为重介悬浮沉降实验,通过一套小型差压装置对悬浮液沉降特性进行研究,获得了34组不同密度、煤泥含量下的沉降曲线;第二部分为重介悬浮液粘度标定实验,实验设备为正弦波振动式粘度计SV10与置顶式搅拌器RW20。
实验室密度测量误差为±0.0083g/cm3;工业现场采用小量程大膜片差压传感器后,密度测量误差为±0 .0037g/cm3,满足课题指标要求。对于粘度,本文给出了两种不同的建模方法,第一种是支持向量机方法,得出了基于直观经验特征提取的“SVR模型”与基于核主成分分析特征提取的“KPCA+SVR模型”;第二种方法是核偏最小二乘方法,得到了“KPLS模型”。这三种模型的粘度测量误差分别为±0.183mPa·s、±0.133 mPa·s与±0.129mPa·s,均满足课题指标要求。采用与粘度测量同样的方法,本文建立了三种煤泥含量测量模型,测量误差分别为3.6%,2.5%与4.0%。
根据本文提出的方法设计的工业样机已经在选煤现场运行。运行结果表明,差压式重介悬浮液密度与粘度一体化测量方法可以在工业上推广应用。
In coal preparation process, density and viscosity are two of the most important parameters of heavy medium suspension which need to be accurately measured. Currently, the isotope densimeter was commonly used to detect the density of heavy medium suspension, but the radioactivity is harmful to people. Furthermore, there is a lack of on-line device to detect the viscosity of heavy medium suspension till now. Most coal preparation plants use a magnetic material content detecting device to measure the magnetic powder content of the suspension firstly, then compute the coal-content of suspension combined with the density detected by isotope densimeter, and take the coal-content as an indicator of suspension viscosity. However, the measurement error of this method is large.
This paper presents an integrated measurement method for density and viscosity of heavy medium suspension based on differential pressure, which achieves density measurement without radioactivity and on-line measurement of viscosity. That method installs a differential pressure sensor on a byroad which is beside the vertical main loop pipeline of heavy medium suspension. When suspension is stability flow in the byroad, the density could be measured according to the output of sensor. While closing the valves of the byroad, heavy medium suspension will start to settle, the viscosity of suspension would be obtained through the settling process.
The purpose of this study is to verify the measurement method presented above through a series of experiments, and to establish the measurement model which measuring viscosity from settling curves. The experiments include two parts. One part is the settling experiments of heavy medium suspension. Through a differential pressure device, 34 groups of settling curves were obtained. The other part is the calibration experiments of suspension’s viscosity employing a viscometer called SV10 and a stirrer named RW20.
The measurement error of density in laboratory is±0.0083g/cm3, and±0.0037 g/cm3 in industry plant due to the smaller range and bigger diaphragm area of sensors. About to viscosity, this paper provides two different ways to establish the measurement model. In the first way on support vector machine, the“SVR model”which extracts features using experience and the“KPCA+SVR model”which extracts features using kernel principle component analysis were obtained. In the other way on kernel partial least squares, the“KPLS model”was build. The measurement errors of these three models are±0.183mPa·s,±0.133mPa·s and±0.129mPa·s respectively. Using the same methods, the coal-content measurement model was established and the measuring errors are 3.6%, 2.5% and 4.0% respectively.
The industrial prototype which designed according to the method proposed by this paper has been running in the coal preparation plant. The results show that the integrated measurement for density and viscosity of heavy medium suspension based on differential pressure can be applied in coal preparation industry.
本文提出一种差压式重介悬浮液密度与粘度一体化测量方案,实现了密度无核测量以及粘度的在线检测。该方案在悬浮液垂直主回路附近开辟一旁路,在旁路上安装一套差压传感器,当悬浮液在旁路中稳定流动时,通过差压输出测量密度;关闭旁路上、下阀门,重介悬浮液在旁路中发生静态沉降,通过沉降过程测量其粘度。
本文的目的就是通过实验对测量方案进行验证,并给出从悬浮液沉降曲线到粘度的建模方法。实验分为两部分,第一部分为重介悬浮沉降实验,通过一套小型差压装置对悬浮液沉降特性进行研究,获得了34组不同密度、煤泥含量下的沉降曲线;第二部分为重介悬浮液粘度标定实验,实验设备为正弦波振动式粘度计SV10与置顶式搅拌器RW20。
实验室密度测量误差为±0.0083g/cm3;工业现场采用小量程大膜片差压传感器后,密度测量误差为±0 .0037g/cm3,满足课题指标要求。对于粘度,本文给出了两种不同的建模方法,第一种是支持向量机方法,得出了基于直观经验特征提取的“SVR模型”与基于核主成分分析特征提取的“KPCA+SVR模型”;第二种方法是核偏最小二乘方法,得到了“KPLS模型”。这三种模型的粘度测量误差分别为±0.183mPa·s、±0.133 mPa·s与±0.129mPa·s,均满足课题指标要求。采用与粘度测量同样的方法,本文建立了三种煤泥含量测量模型,测量误差分别为3.6%,2.5%与4.0%。
根据本文提出的方法设计的工业样机已经在选煤现场运行。运行结果表明,差压式重介悬浮液密度与粘度一体化测量方法可以在工业上推广应用。
In coal preparation process, density and viscosity are two of the most important parameters of heavy medium suspension which need to be accurately measured. Currently, the isotope densimeter was commonly used to detect the density of heavy medium suspension, but the radioactivity is harmful to people. Furthermore, there is a lack of on-line device to detect the viscosity of heavy medium suspension till now. Most coal preparation plants use a magnetic material content detecting device to measure the magnetic powder content of the suspension firstly, then compute the coal-content of suspension combined with the density detected by isotope densimeter, and take the coal-content as an indicator of suspension viscosity. However, the measurement error of this method is large.
This paper presents an integrated measurement method for density and viscosity of heavy medium suspension based on differential pressure, which achieves density measurement without radioactivity and on-line measurement of viscosity. That method installs a differential pressure sensor on a byroad which is beside the vertical main loop pipeline of heavy medium suspension. When suspension is stability flow in the byroad, the density could be measured according to the output of sensor. While closing the valves of the byroad, heavy medium suspension will start to settle, the viscosity of suspension would be obtained through the settling process.
The purpose of this study is to verify the measurement method presented above through a series of experiments, and to establish the measurement model which measuring viscosity from settling curves. The experiments include two parts. One part is the settling experiments of heavy medium suspension. Through a differential pressure device, 34 groups of settling curves were obtained. The other part is the calibration experiments of suspension’s viscosity employing a viscometer called SV10 and a stirrer named RW20.
The measurement error of density in laboratory is±0.0083g/cm3, and±0.0037 g/cm3 in industry plant due to the smaller range and bigger diaphragm area of sensors. About to viscosity, this paper provides two different ways to establish the measurement model. In the first way on support vector machine, the“SVR model”which extracts features using experience and the“KPCA+SVR model”which extracts features using kernel principle component analysis were obtained. In the other way on kernel partial least squares, the“KPLS model”was build. The measurement errors of these three models are±0.183mPa·s,±0.133mPa·s and±0.129mPa·s respectively. Using the same methods, the coal-content measurement model was established and the measuring errors are 3.6%, 2.5% and 4.0% respectively.
The industrial prototype which designed according to the method proposed by this paper has been running in the coal preparation plant. The results show that the integrated measurement for density and viscosity of heavy medium suspension based on differential pressure can be applied in coal preparation industry.
引文
[1]顾文卿.新编选煤实用手册.北京:中国煤炭工业出版社,2005. 317~404
[2]吴庆勇.重介悬浮液固相体积浓度的计算.科技资讯,2006,24:27
[3]祁泽民,符东旭.选煤重介质悬浮液稳定性分析.煤炭科学技术,2008,36(6):107~109
[4]吕霞付,蔡绍皙.液体密度动态测量用传感器.传感器世界,1998,4:8~14
[5]李存蔚,荣焕成,孙林兔.重介密度自动控制系统研究.选煤技术,2002,4:45~47
[6] M S Greenwood, J R Skorpik, J A Bamberger, et al. On-line ultrasonic density for process control of liquids and slurries. Ultrasonics. 1999, 37:159~171
[7] Judith A Bamberger, Margaret S Greenwood. Measuring fluid and slurry density and solids concentration non-invasively. Ultrasonics. 2004, 42:563~567
[8]郭秀清,杨兴国.重介悬浮液密度及液位的检测与自动控制.选煤技术,1999,6:40~42
[9]赵能富.用同位素密度计测定复选煤浆浓度时的精度.选煤技术,1992, 5:20~24
[10] S K Kawatra, A K Bakshi. On-line measurement of viscosity and determination of flow types for mineral suspensions. Int.J.Miner.Process. 1996(47) :275~283
[11] J D Grobler, R F Sandenbergh, P C Pistorius. The stability of ferrosilicon dense medium suspensions. The Journal of The South African Institute of Ming and Metallurgy. 2002: 83~86
[12] M S Greenwood, J A Bamberger. Measurement of viscosity and shear wave velocity of liquid. Ultrasonics. 2002,39: 623~630
[13] Margaret S Greenwood, Justus D Adamson, Leonard J Bond. Measurement of the viscosity-density product using multiple reflections of ultrasonic shear horizontal waves. Ultrasonics. 2006, 44:1031~1036
[14]赵春祥.重介质悬浮液特性控制问题的探讨.选煤技术,2000,4:17~19
[15]柯元明,李明,李兰.流动液体密度粘度连续测量方案—压差式密度粘度测量系统.天然气与石油,2002,3(20):52~54
[16]丁宏大,刘德忠.矿浆静态自然沉降实验研究.有色金属,1985,2:44~50
[17]彭坚,黄铭荣.悬浮液静置沉降试验方法的研究.给水排水,1985,05:2~10
[18]谢洪勇,张大为,张礼鸣.非球形颗粒及颗粒的自由沉降速度.中国粉体技术,2005,11:39~41
[19] Y B He, J S Laskowski, B Klein. Particle movement in non-Newtonian slurries: the effect of yield stress on dense medium separation. Chemical Engineering Science. 2001,56: 2991~2998
[20] A Rushton, A S Ward, R G Holdich.固液两相过滤及分离技术(朱企新,许莉,谭蔚等).北京:化学工业出版社,2005.56~155
[21] V N Vapnik. The Nature of Statistical Learning Theory. New York: Springer-Verlag, 1995.
[22]张学工.关于统计学校理论与支持向量机.自动化学报,2000,26(1):32~42
[23] Alex J Smola, Bernhard Scholkopf. A tutorial on Support vector regression. Royal Holloway College, London, U.K., NeuroCOLT Tech. Rep. TR, 1998
[24] Yunqiu Wu, Zongxi Tang, Yuehang Xu, et al. Support Vector Regression for Measuring Electromagnetic Parameters of Magnetic Thin-Film Materials. IEEE transactions on magnetics. 2007, 43(12):4071~4075
[25] L J Cao, Francis E H Tay. Support Vector Machine With Adaptive Parameters in Financial Time Series Forecasting. IEEE transaction on neural networks. 2003,14(6):1506~1518
[26]邓乃杨,田英杰.数据挖掘中的新方法——支持向量机.北京:科学出版社,2004.223~274
[27] Wenjian Wang, Zongben Xu, Weizhen Lu, et al. Determination of the spread parameter in the Gaussian kernel for classification and regression. Neurocomputing. 2003,55: 643~663
[28] Momma M, Bennett K P. A pattern search method for model selection of support vector regression. In. Proc. of SIAM Conference on Data Mining. 2002
[29] Yulin Dong, Zhonghang Xia, Manghui Tu, et al. An Optimization Method for Selecting Parameters in Support Vector Machines. Sixth International Conference on Machine Learning and Applications. 2007:1~6
[30]李良敏,温广瑞,王生昌.基于遗传算法的回归型支持向量机参数选择.计算机工程与应用,2008,44(7):23~26
[31] Olivier Vapnik, Vladimir Vapnik, Olivier Bousquet, et al. Choose Multiple Parameters for Support Vector Machines. Machine Learning. 2005,46:131~159
[32] Gautam Kunapuli, Kristin P ennett, Jing Hu, et al. Bilevel Model Selection for Support Vector Machines. Centre de Recherches Mathematiques CRM Proceedings and Lecture Notes. 2008, 45:129~158
[33] C C Chang, C J Lin. LIBSVM: a library for support vector machines (version 2.88). 2009. [Online], Available: http://www.csie.ntu.edu.tw/~cjlin/libsvm/
[34]王惠文,吴载斌,孟浩.偏最小二乘回归的线性与非线性方法.北京:国防工业出版社,2006.55~123
[35] Roman Rosipal, Leonard J Trejo. Kernel Partial Least Squares Regression in Reproducing Kernel Hilbert Space. Journal of Machine Learning Research. 2001, 2: 97~123
[36]韦振中.基于核主成分分析的特征提取方法.广西工学院学报,2006,17(4): 27~31
[37] Cao L J, Chua K S, Chong W K. A comparison of PCA, KPCA and ICA for dimensionality reduction in support vector machine. Neurocomputing. 2003, 55(2):321~336
[38]郝凤印.选煤手册.北京:煤炭工业出版社,1993.263~273
[2]吴庆勇.重介悬浮液固相体积浓度的计算.科技资讯,2006,24:27
[3]祁泽民,符东旭.选煤重介质悬浮液稳定性分析.煤炭科学技术,2008,36(6):107~109
[4]吕霞付,蔡绍皙.液体密度动态测量用传感器.传感器世界,1998,4:8~14
[5]李存蔚,荣焕成,孙林兔.重介密度自动控制系统研究.选煤技术,2002,4:45~47
[6] M S Greenwood, J R Skorpik, J A Bamberger, et al. On-line ultrasonic density for process control of liquids and slurries. Ultrasonics. 1999, 37:159~171
[7] Judith A Bamberger, Margaret S Greenwood. Measuring fluid and slurry density and solids concentration non-invasively. Ultrasonics. 2004, 42:563~567
[8]郭秀清,杨兴国.重介悬浮液密度及液位的检测与自动控制.选煤技术,1999,6:40~42
[9]赵能富.用同位素密度计测定复选煤浆浓度时的精度.选煤技术,1992, 5:20~24
[10] S K Kawatra, A K Bakshi. On-line measurement of viscosity and determination of flow types for mineral suspensions. Int.J.Miner.Process. 1996(47) :275~283
[11] J D Grobler, R F Sandenbergh, P C Pistorius. The stability of ferrosilicon dense medium suspensions. The Journal of The South African Institute of Ming and Metallurgy. 2002: 83~86
[12] M S Greenwood, J A Bamberger. Measurement of viscosity and shear wave velocity of liquid. Ultrasonics. 2002,39: 623~630
[13] Margaret S Greenwood, Justus D Adamson, Leonard J Bond. Measurement of the viscosity-density product using multiple reflections of ultrasonic shear horizontal waves. Ultrasonics. 2006, 44:1031~1036
[14]赵春祥.重介质悬浮液特性控制问题的探讨.选煤技术,2000,4:17~19
[15]柯元明,李明,李兰.流动液体密度粘度连续测量方案—压差式密度粘度测量系统.天然气与石油,2002,3(20):52~54
[16]丁宏大,刘德忠.矿浆静态自然沉降实验研究.有色金属,1985,2:44~50
[17]彭坚,黄铭荣.悬浮液静置沉降试验方法的研究.给水排水,1985,05:2~10
[18]谢洪勇,张大为,张礼鸣.非球形颗粒及颗粒的自由沉降速度.中国粉体技术,2005,11:39~41
[19] Y B He, J S Laskowski, B Klein. Particle movement in non-Newtonian slurries: the effect of yield stress on dense medium separation. Chemical Engineering Science. 2001,56: 2991~2998
[20] A Rushton, A S Ward, R G Holdich.固液两相过滤及分离技术(朱企新,许莉,谭蔚等).北京:化学工业出版社,2005.56~155
[21] V N Vapnik. The Nature of Statistical Learning Theory. New York: Springer-Verlag, 1995.
[22]张学工.关于统计学校理论与支持向量机.自动化学报,2000,26(1):32~42
[23] Alex J Smola, Bernhard Scholkopf. A tutorial on Support vector regression. Royal Holloway College, London, U.K., NeuroCOLT Tech. Rep. TR, 1998
[24] Yunqiu Wu, Zongxi Tang, Yuehang Xu, et al. Support Vector Regression for Measuring Electromagnetic Parameters of Magnetic Thin-Film Materials. IEEE transactions on magnetics. 2007, 43(12):4071~4075
[25] L J Cao, Francis E H Tay. Support Vector Machine With Adaptive Parameters in Financial Time Series Forecasting. IEEE transaction on neural networks. 2003,14(6):1506~1518
[26]邓乃杨,田英杰.数据挖掘中的新方法——支持向量机.北京:科学出版社,2004.223~274
[27] Wenjian Wang, Zongben Xu, Weizhen Lu, et al. Determination of the spread parameter in the Gaussian kernel for classification and regression. Neurocomputing. 2003,55: 643~663
[28] Momma M, Bennett K P. A pattern search method for model selection of support vector regression. In. Proc. of SIAM Conference on Data Mining. 2002
[29] Yulin Dong, Zhonghang Xia, Manghui Tu, et al. An Optimization Method for Selecting Parameters in Support Vector Machines. Sixth International Conference on Machine Learning and Applications. 2007:1~6
[30]李良敏,温广瑞,王生昌.基于遗传算法的回归型支持向量机参数选择.计算机工程与应用,2008,44(7):23~26
[31] Olivier Vapnik, Vladimir Vapnik, Olivier Bousquet, et al. Choose Multiple Parameters for Support Vector Machines. Machine Learning. 2005,46:131~159
[32] Gautam Kunapuli, Kristin P ennett, Jing Hu, et al. Bilevel Model Selection for Support Vector Machines. Centre de Recherches Mathematiques CRM Proceedings and Lecture Notes. 2008, 45:129~158
[33] C C Chang, C J Lin. LIBSVM: a library for support vector machines (version 2.88). 2009. [Online], Available: http://www.csie.ntu.edu.tw/~cjlin/libsvm/
[34]王惠文,吴载斌,孟浩.偏最小二乘回归的线性与非线性方法.北京:国防工业出版社,2006.55~123
[35] Roman Rosipal, Leonard J Trejo. Kernel Partial Least Squares Regression in Reproducing Kernel Hilbert Space. Journal of Machine Learning Research. 2001, 2: 97~123
[36]韦振中.基于核主成分分析的特征提取方法.广西工学院学报,2006,17(4): 27~31
[37] Cao L J, Chua K S, Chong W K. A comparison of PCA, KPCA and ICA for dimensionality reduction in support vector machine. Neurocomputing. 2003, 55(2):321~336
[38]郝凤印.选煤手册.北京:煤炭工业出版社,1993.263~273