基于组合神经网络的软测量技术研究
|
摘要
熔融指数是聚丙烯生产过程中最重要的质量指标。由于工艺设备以及技术上的原因,熔融指数难以在线测量,从现场采集聚丙烯产品样品进行化验则会造成测量滞后的存在,导致无法对熔融指数进行有效在线控制。针对聚丙烯生产过程熔融指数测量存在的难题,提出一种组合神经网络软测量建模方法,并用于建立熔融指数的软测量模型。
本文的主要研究内容如下:
(1)首先简单介绍了聚合过程的特点,以及聚合反应过程研究存在的问题。探讨了软测量技术在聚合反应过程应用的可行性和意义。介绍了软测量技术的基本思想和实现步骤,并对各种软测量建模方法及其在工业领域中的应用现状进行了综述,并指出软测量技术在处理强非线性和不确定性对象时还有待完善。
(2)具体分析了常用的BP神经网络算法,包括BP网络结构和BP训练过程及算法。针对人工神经网络在建立非线性对象模型时的缺点,引入组合预测方法。对组合预测的研究现状、分类、组合系数的确定进行了详细的介绍。最后将神经网络与组合预测方法相结合提出了组合神经网络,并将最小化最大绝对误差作为求解组合权重的准则,提出了基于最小化最大绝对误差的组合神经网络建模方法。
(3)综合分析Spheripol工艺的催化剂体系、均聚反应机理和生产工艺流程,确定影响熔融指数的辅助变量。其次,对选取的样本数据进行预处理,包括异常数据剔除、随机误差处理和数据归一化。随后采用经验法和凑试法相结合,确定了BP网络隐含层的数目和节点数。最后采用组合神经网络建模,并将组合神经网络同单一BP神经网络进行性能比较,通过比较分析说明了组合神经网络的有效性。
Melt index is the most important indicator of quality in the production process of polypropylene. Because of the constraints of process equipment and technical, the melt index is difficult to measure online. If we directly measure the melt index of the polypropylene product samples, that will result the measurement delay, melt index can not be controlled effectively online. Because the melt index is difficult to measure online, model of melt index prediction based on stacked neural networks was established.
This article mainly includes the following:
(1) Above all, introduced the characteristics of the polymerization process and problem of polymerization process study. Discussed the feasibility and significance of soft sensor. Introduced the basic idea of soft sensor and its implementation steps, generally described the situation of various soft sensor modeling method and its application in the industrial field.
(2) Analyzed some commonly artificial neural network algorithm, including the RBF neural network and BP neural network. Because of the defect of establishing a nonlinear model base on artificial neural network, introduced combined forecasting method. Detailedly described the status quo, classification and combination coefficient of combined forecasting. Finally, proposed stacked neural networks based on neural network and combined forecasting methods, and to minimizing the maximum absolute error as the criteria for solving the combination weight. Proposed the combined neural network modeling based on minimizing the maximum absolute error.
(3) Integrated analysed the catalyst system of Spheripol process、homopolymerization mechanism and production process, determine auxiliary variables of impacting melt index. Secondly, preprocessed sample data, including different date exclusion, random error processing and data normalization. Then, determined the number of BP network hidden layer and nodes based on experience piece-try method. Finally, build combined neural network model, and through comparing combined neural network with a single BP neural network, proved the effectiveness of combined neural network.
本文的主要研究内容如下:
(1)首先简单介绍了聚合过程的特点,以及聚合反应过程研究存在的问题。探讨了软测量技术在聚合反应过程应用的可行性和意义。介绍了软测量技术的基本思想和实现步骤,并对各种软测量建模方法及其在工业领域中的应用现状进行了综述,并指出软测量技术在处理强非线性和不确定性对象时还有待完善。
(2)具体分析了常用的BP神经网络算法,包括BP网络结构和BP训练过程及算法。针对人工神经网络在建立非线性对象模型时的缺点,引入组合预测方法。对组合预测的研究现状、分类、组合系数的确定进行了详细的介绍。最后将神经网络与组合预测方法相结合提出了组合神经网络,并将最小化最大绝对误差作为求解组合权重的准则,提出了基于最小化最大绝对误差的组合神经网络建模方法。
(3)综合分析Spheripol工艺的催化剂体系、均聚反应机理和生产工艺流程,确定影响熔融指数的辅助变量。其次,对选取的样本数据进行预处理,包括异常数据剔除、随机误差处理和数据归一化。随后采用经验法和凑试法相结合,确定了BP网络隐含层的数目和节点数。最后采用组合神经网络建模,并将组合神经网络同单一BP神经网络进行性能比较,通过比较分析说明了组合神经网络的有效性。
Melt index is the most important indicator of quality in the production process of polypropylene. Because of the constraints of process equipment and technical, the melt index is difficult to measure online. If we directly measure the melt index of the polypropylene product samples, that will result the measurement delay, melt index can not be controlled effectively online. Because the melt index is difficult to measure online, model of melt index prediction based on stacked neural networks was established.
This article mainly includes the following:
(1) Above all, introduced the characteristics of the polymerization process and problem of polymerization process study. Discussed the feasibility and significance of soft sensor. Introduced the basic idea of soft sensor and its implementation steps, generally described the situation of various soft sensor modeling method and its application in the industrial field.
(2) Analyzed some commonly artificial neural network algorithm, including the RBF neural network and BP neural network. Because of the defect of establishing a nonlinear model base on artificial neural network, introduced combined forecasting method. Detailedly described the status quo, classification and combination coefficient of combined forecasting. Finally, proposed stacked neural networks based on neural network and combined forecasting methods, and to minimizing the maximum absolute error as the criteria for solving the combination weight. Proposed the combined neural network modeling based on minimizing the maximum absolute error.
(3) Integrated analysed the catalyst system of Spheripol process、homopolymerization mechanism and production process, determine auxiliary variables of impacting melt index. Secondly, preprocessed sample data, including different date exclusion, random error processing and data normalization. Then, determined the number of BP network hidden layer and nodes based on experience piece-try method. Finally, build combined neural network model, and through comparing combined neural network with a single BP neural network, proved the effectiveness of combined neural network.
引文
[1] Mcacvoy T J. Conternplative stance for chemical process control[J]. Automatica. 1922, 28(2): 441-442.
[2]蒋慰孙,蒋敏伟.过程控制21实际展望[J].世界仪表和自动化. 1999, 3(4): 10-14.
[3] Yi H S, Kin J H, Han C, et al. Plantwide optimal grade transition for an industrial high-density polyethylene plant [J]. Ind Eng Chem Res, 2003, 42(1): 91-98.
[4] Flores T A, Biegler L T, Saldivar G E. Optimal grade transitions in the high-impact polystyrene polymerization process[J]. Ind Eng Chem Res, 2006, 45(18): 6175-6189.
[5] HanI-S,Han C, Chung C-B. [J]. J. Appl. Poly. Sci. 2005, 95: 967–974.
[6] Ohshima M, Tanigaki M. [J]. J. Processc Control, 2000, 10: 135-148.
[7] Feil B, Abonyi J, Pach P, Nemeth S, Arva P, Nemeth M, Nagy G. Semi–mechanistic models for state-estimation soft sensor for polymer melt index prediction[C]. 7th International Conference on Artificial Intelligence and Soft Computing, 2004, 1111-1117.
[8] Hinchliffe M, Montague G, Willis M, et al. [J]. AIChE J. 2003, 49, 3127-3137.
[9]金以慧,王桂增.工程控制的发展和展望[J].控制理论与应用. 1997, 14(2): 145-151.
[10] Han I S, Han C, Chung C B. Melt index modeling with support vector machines, partial least squares, and artificial neural networks [J]. J Appl Polym Sci, 2005, 95(4): 967-974.
[11] Shi J, Liu X G, Sun Y X. Melt index prediction by neural networks based on independent component analysis and multi-scale analysis[J]. Neurocomputing, 2006, 70(1-3): 280-287.
[12] Zhang J, Qi B X, Yong M. Inferential estimation of polymer melt index usingsequentially trained bootstrap aggregated neutral networks[J]. Chem Eng Technol, 2006, 29(4): 442-448.
[13]古勇,苏宏业,王朝晖等. FCCU软测量综述[J].化工自动化及仪表. 1998,增刊: 6-10.
[14]张明君,皮道应,孙优贤.基于工程观点的软仪表开发策略[J].化工自动化及仪表. 1996, 23(6): 34-36.
[15]骆晨钟,邵惠鹤.软仪表技术及其工业应用[J].仪表技术与传感器. 1999, 6(1): 32-39.
[16]徐敏,俞金寿.软测量技术[J],石油化工自动化. 1998, 6(2): 1-3.
[17]荣冈,金晓明,王树青.先进控制技术及应用[J].化工自动化及仪表. 1999, 26(4): 70-72.
[18]孙欣,王金春,何声亮.过程软测量[J].自动化仪表. 1995, 16(8): 1-5.
[19] Van den Bos A. Application of statistical parameter estimation methods to physical measurement[J]. J Phys E: Sci Instrum. 1977, 10: 753-760.
[20] Riviere J-M. Bayart M, Thiriet J-M. Intelligent instruments: some modeling approaches[J]. Measurement + Control. 1996, 29(4): 179-186.
[21] van den Bos A and Eykboff P. Model building and parameter estimation as means for intelligent measurement[J]. Measurement. 1988, 6(1): 25-32.
[22]蒙建波,朱林章.检测理论的现状及发展展望[J].化工自动化及仪表. 1994, 31(3): 40-41, 45.
[23]赵新民,陈海军.模型化测量的发展现状与展望[J].计量学报. 1992, 13(4): 314-319.
[24]骆晨钟,邵惠鹤.基于机理分析的方法实现催化裂化反应预测转化率、产品产率的软测量[J].化工自动化及仪表, 1999, 26(2): 29-33.
[25]赵征.基于信息融合的锅炉燃烧状态参数检测技术研究[J].华北电力学, 2007, 35(8): 65-70.
[26]赵科,王生铁,张计科.三容水箱的机理建模[J].控制工程, 2006, 13(6): 521-525.
[27] Weber R and Brosilow C. The use of secondary measurements to improve control[J]. AIChE J. 1972, 18(3): 614-623.
[28] Lee D S, Vanrolleghem P A. Adaptive consensus principal component analysis for on-line batch process monitoring [J]. Environmental Monitoring and Assessment, 2004, 92(1): 119-135.
[29] Zhao X, Wen X J, Shao H H. Combination method of principal component analysis and support vector machine for on-line process monitoring and fault diagnosis [J]. Journal of Donghua University (English Edition), 2006, 23(1): 53-58.
[30]翟军勇,费树珉.软测量技术和约束控制在精馏塔优化控制中的应用[J].工业仪表与自动化装置, 2004, 24(2): 24-26.
[31]王近邻,邵之江,张仲广.乙醛生产过程中的软测量实现[J].化工自动化及仪表,2004,31(1):56-59.
[32]梁军,汪小勇,王文庆.基于神经网络PLS方法的软测量建模研究[J].浙江大学学报, 2004, 38(6): 676-681.
[33]朱国强,刘士荣,俞金寿.基于支持向量机的数据建模在软测量建模中的应用[J].华东理工大学学报. 2002, 22(9): 6-11.
[34]张英,苏宏业,褚健.基于模糊最小二乘支持向量机的软测量建模[J].控制与决策. 2005, 20(6): 621-624.
[35]阎威武,常俊林,邵惠鹤.基于滚动时间窗的最小二乘支持向量机回归估计方法及仿真[J].上海交通大学学报, 2004, 38(4): 524-526
[36]催万照,朱长纯,保文星.最小二乘小波支持向量机载非线性系统辨识中的应用[J].西安交通大学学报, 2004, 38(6): 562-565.
[37] Keerthi S S, Shevade S K, Bhattacharyya C, et al. A fast iterative nearest point algorithm for support vector machine classifier design[J].IEEE Transactions on Neural Networks, 2000,11(1): 124-136.
[38] Yang M H, Ahuja N. Ageometric approach to train support vector machines[A].Proceedings of CVPR 2000[C]. Hilton Head Island,2000. 430-437.
[39] B.Boser, I.Guyon ,V.Vapnik. A T raining Algorithm for Optimal Margin Classifiers [A]. Proceedings of the 5th Annual ACM Work shop on Computational L earning Theory[C ]. ACM Press, 1992.
[40] Lin Chun-Fu, Wang Sheng-De. Fuzzy support vector machines [J]. IEEETransactions on Neural Networks, 2002, 13(2): 464-471.
[41] Suykens J A K, Vandewalle J. Least square support vector machines classifiers[J]. Neural Network Letters, 1999, 19(3):293-300.
[42]张学工.关于统计学习理论与支持矢量机[J].自动化学报,2000 , 26 (1).32-42.
[43]冯瑞,张浩然,邵惠鹤.基于SVM的软测量建模[J].信息与控制,2002,31(6):567-5711.
[44]王定成,方廷健,高理富.支持向量机回归在线建模与应用[J].控制与决策,2003,18(1):89-931.
[45]马勇,黄德先,金以慧.基于支持向量机的软测量建模方法[J].信息与控制,2004,33(4):120-125.
[46]阎威武,朱宏栋,邵惠鹤.基于最小二乘支持向量机的软测量建模[J].系统仿真学报,2003,15(10):1494-1496.
[47]王旭东,邵惠鹤.基于神经网络的通用软测量技术[J].自动化学. 1998, 24(5): 702-706.
[48] Shaw A.M., Doyle III F.J. Multivariable nonlinear control applications for a high purity distillation column using a recurrent dynamic neuron model[J]. J.Proc. Cont. .1997, 7 (4) :255-262.
[49] Joseph B and Brosilow C B. Inferential control of processes: Part I Steady state analysis and design[J]. AIChE J. 1978, 15(2): 150-159.
[50]张磊,胡春,钱锋.BP改进算法及其在乙二醇精制软测量中的应用[J].自动化仪表,2005,26(6):31-34.
[51]杨凌,刘进军,金强.基于改进RBF网络的乙烯纯度软测量建模方法[J].微计算机信息,2008,24(10):158-200.
[52]黄江平,陶慧汇,朱志高.基于BP网络的醋酸乙烯聚合率软测量研究[J].机床与液压,2009,37(8):286-287.
[53]庄利锋,杨慧中,张志强.一种分布式模糊神经网络在熔融指数软测量中的应用[J].化工自动化及仪表,2004,31(4):16-18.
[54] Fikri Akdeniz. The Examination And Analysis of Residuals For Some Biased Estimatorsin Linear Regression[J]. Commun. Statist. Theory Meth , 2001,30(6),1171-1178.
[55]章卫国,杨向忠.模糊控制理论与应用[M].西安:西北工业大学出版社, 1990: 34-58.
[56]刘瑞兰,苏宏业,褚健.基于改进模糊神经网络的软测量建模方法[J].信息与控制, 2003, 32(4): 367-370.
[57]田华阁,车荣杰,王平.基于机制建模与模糊建模相结合的聚丙烯熔融指数软测量方法[J].中国石油大学学报, 2009, 33(3): 162-165.
[58] Kohonen T. Self-Organizing Maps[M]. Berlin: Springer-Verlag, 1997, 50-75.
[59] Kosko B. Neural networks and fuzzy systems[M]. Upper Saddle River :Prentice Hall, Inc, 1992, 32-45.
[60] Mhasker H N. Neural networks for optimal approximation of smooth and analytic functions[J]. Neural Computation, 1996, 32(8): 1731-1742.
[61] Poggio T, Girosi F. Regilarization algorithms for learning that are equivalent to multilayer networks[J]. Science, 1990, 1247(32): 978-982.
[62]史忠植.神经网络[M].北京:高等教育出版社, 2009, 48-63.
[63] Haykin S. Adaptive Filter Theory. 3rd. Upper Saddle River[M]. NJ: Prentice Hall, 1996, 56-89.
[64] Hornik K, Stinchcombe M, White H. Multilayer feedforward networks are universal approximators[J]. Neural Networks, 1989, 2(5): 359-366.
[65] Naguyen D and Widrow B. Improving the learning speed of 2-Layer Neural Networks by Choosing Initial Values of Adaptive Weights[C]. Proceedings of the International Joint Conference On Neural Networks. San Diego, CA,1990(3): 21-26.
[66] Dayan P. Recurrent sampling models for the Helmholtz machine[J]. Neural Computational 1999, 11(9): 653-677.
[67]傅庚,唐小我,曾勇.广义递归方差倒数组合预测方法研究[J].电子科技大学学报, 1995, 24(2): 211-217.
[68]陈友华.基于预测有效度的组合预测模型研究[J].预测, 1997, 16(3): 72-73.
[69]陈友华,侯定丕.基于预测有效度的优性组合预测模型研究[J].中国科技大学学报, 2002, 32(2): 172-180.
[70]王应明.基于相关性的组合预测方法研究[J].预测, 2002, 21(2): 58-62.
[71]王应明,罗英.调和平均组合预测中的参数估计技术[J].系统工程与电子技术, 1997, 19(10): 45-49.
[72]王应明.广义加权几何平均组合预测方法研究[J].厦门大学学报(自然科学版), 1998, 37(1): 6-10.
[73] Wolpert D H. Stacked generalization [J]. Neural Networks, 1992, 5(2): 241-259.
[74] Sridhar D V, Bartlett E B, Seagrave R C. Information theoretic approach for combining neural network process models[J]. Neural Networks, 1999,12(6): 915-926.
[75] Zhang J. Developing robust non-linear models through bootstrap aggregated neural networks [J]. Neurocomputing, 1999, 25(1): 93-113.
[76] Sridhar D V,Seagrave R C, Bartlett E B. Process modelling using stacked neural networks[J ]. AIChE J, 1996, 42(9): 2529-2539.
[77] Hashem S. Optimal linear combinations of neural networks[J]. Neural Networks, 1997, 10(4): 599-614.
[78] Zhang J, Morris A J, Martin E B, et al. Prediction of polymer quality in batch polymerization reactors using robust neural networks[J]. Chem Eng J , 1998, 69(2): 135-143.
[79]钟国平.浅述我国聚丙烯工业发展[J].石油化工, 1996, 25(10): 739-744.
[80]杨维洪.聚丙烯发展近况[J].石油化工, 2001, 30(4): 319-324.
[81]黄葆同,陈伟主编,茂金属催化剂及其烯烃版合物[M].北京:化学工业出版社, 2000, 134-145.
[82]洪定一主编,聚丙烯——原理、工艺和技术[M],北京:中国石化出版社, 2002, 370-377.
[83]吕砚山,赵正琦. BP神经网络的优化及应用研究[J].北京化工大学学报, 2001, 31(1): 67-69.
[84]张德贤.前向神经网络合理隐含层结点个数估计[J].计算机工程与应用, 2005, 30(5): 21-24.
[85] Franco Scarselli, Ah Chung Tsoi. Universal approximation using feed-forward neural networks: A survey of some existing methods and some new results[J].Neural Networks. 1998, 11(1): 15-37.
[86] S V Kamarthi, S Pittner. Accelerating neural network training using weight extrapolation[J]. Neural networks. 1999, 12(12): 1285-1299.
[2]蒋慰孙,蒋敏伟.过程控制21实际展望[J].世界仪表和自动化. 1999, 3(4): 10-14.
[3] Yi H S, Kin J H, Han C, et al. Plantwide optimal grade transition for an industrial high-density polyethylene plant [J]. Ind Eng Chem Res, 2003, 42(1): 91-98.
[4] Flores T A, Biegler L T, Saldivar G E. Optimal grade transitions in the high-impact polystyrene polymerization process[J]. Ind Eng Chem Res, 2006, 45(18): 6175-6189.
[5] HanI-S,Han C, Chung C-B. [J]. J. Appl. Poly. Sci. 2005, 95: 967–974.
[6] Ohshima M, Tanigaki M. [J]. J. Processc Control, 2000, 10: 135-148.
[7] Feil B, Abonyi J, Pach P, Nemeth S, Arva P, Nemeth M, Nagy G. Semi–mechanistic models for state-estimation soft sensor for polymer melt index prediction[C]. 7th International Conference on Artificial Intelligence and Soft Computing, 2004, 1111-1117.
[8] Hinchliffe M, Montague G, Willis M, et al. [J]. AIChE J. 2003, 49, 3127-3137.
[9]金以慧,王桂增.工程控制的发展和展望[J].控制理论与应用. 1997, 14(2): 145-151.
[10] Han I S, Han C, Chung C B. Melt index modeling with support vector machines, partial least squares, and artificial neural networks [J]. J Appl Polym Sci, 2005, 95(4): 967-974.
[11] Shi J, Liu X G, Sun Y X. Melt index prediction by neural networks based on independent component analysis and multi-scale analysis[J]. Neurocomputing, 2006, 70(1-3): 280-287.
[12] Zhang J, Qi B X, Yong M. Inferential estimation of polymer melt index usingsequentially trained bootstrap aggregated neutral networks[J]. Chem Eng Technol, 2006, 29(4): 442-448.
[13]古勇,苏宏业,王朝晖等. FCCU软测量综述[J].化工自动化及仪表. 1998,增刊: 6-10.
[14]张明君,皮道应,孙优贤.基于工程观点的软仪表开发策略[J].化工自动化及仪表. 1996, 23(6): 34-36.
[15]骆晨钟,邵惠鹤.软仪表技术及其工业应用[J].仪表技术与传感器. 1999, 6(1): 32-39.
[16]徐敏,俞金寿.软测量技术[J],石油化工自动化. 1998, 6(2): 1-3.
[17]荣冈,金晓明,王树青.先进控制技术及应用[J].化工自动化及仪表. 1999, 26(4): 70-72.
[18]孙欣,王金春,何声亮.过程软测量[J].自动化仪表. 1995, 16(8): 1-5.
[19] Van den Bos A. Application of statistical parameter estimation methods to physical measurement[J]. J Phys E: Sci Instrum. 1977, 10: 753-760.
[20] Riviere J-M. Bayart M, Thiriet J-M. Intelligent instruments: some modeling approaches[J]. Measurement + Control. 1996, 29(4): 179-186.
[21] van den Bos A and Eykboff P. Model building and parameter estimation as means for intelligent measurement[J]. Measurement. 1988, 6(1): 25-32.
[22]蒙建波,朱林章.检测理论的现状及发展展望[J].化工自动化及仪表. 1994, 31(3): 40-41, 45.
[23]赵新民,陈海军.模型化测量的发展现状与展望[J].计量学报. 1992, 13(4): 314-319.
[24]骆晨钟,邵惠鹤.基于机理分析的方法实现催化裂化反应预测转化率、产品产率的软测量[J].化工自动化及仪表, 1999, 26(2): 29-33.
[25]赵征.基于信息融合的锅炉燃烧状态参数检测技术研究[J].华北电力学, 2007, 35(8): 65-70.
[26]赵科,王生铁,张计科.三容水箱的机理建模[J].控制工程, 2006, 13(6): 521-525.
[27] Weber R and Brosilow C. The use of secondary measurements to improve control[J]. AIChE J. 1972, 18(3): 614-623.
[28] Lee D S, Vanrolleghem P A. Adaptive consensus principal component analysis for on-line batch process monitoring [J]. Environmental Monitoring and Assessment, 2004, 92(1): 119-135.
[29] Zhao X, Wen X J, Shao H H. Combination method of principal component analysis and support vector machine for on-line process monitoring and fault diagnosis [J]. Journal of Donghua University (English Edition), 2006, 23(1): 53-58.
[30]翟军勇,费树珉.软测量技术和约束控制在精馏塔优化控制中的应用[J].工业仪表与自动化装置, 2004, 24(2): 24-26.
[31]王近邻,邵之江,张仲广.乙醛生产过程中的软测量实现[J].化工自动化及仪表,2004,31(1):56-59.
[32]梁军,汪小勇,王文庆.基于神经网络PLS方法的软测量建模研究[J].浙江大学学报, 2004, 38(6): 676-681.
[33]朱国强,刘士荣,俞金寿.基于支持向量机的数据建模在软测量建模中的应用[J].华东理工大学学报. 2002, 22(9): 6-11.
[34]张英,苏宏业,褚健.基于模糊最小二乘支持向量机的软测量建模[J].控制与决策. 2005, 20(6): 621-624.
[35]阎威武,常俊林,邵惠鹤.基于滚动时间窗的最小二乘支持向量机回归估计方法及仿真[J].上海交通大学学报, 2004, 38(4): 524-526
[36]催万照,朱长纯,保文星.最小二乘小波支持向量机载非线性系统辨识中的应用[J].西安交通大学学报, 2004, 38(6): 562-565.
[37] Keerthi S S, Shevade S K, Bhattacharyya C, et al. A fast iterative nearest point algorithm for support vector machine classifier design[J].IEEE Transactions on Neural Networks, 2000,11(1): 124-136.
[38] Yang M H, Ahuja N. Ageometric approach to train support vector machines[A].Proceedings of CVPR 2000[C]. Hilton Head Island,2000. 430-437.
[39] B.Boser, I.Guyon ,V.Vapnik. A T raining Algorithm for Optimal Margin Classifiers [A]. Proceedings of the 5th Annual ACM Work shop on Computational L earning Theory[C ]. ACM Press, 1992.
[40] Lin Chun-Fu, Wang Sheng-De. Fuzzy support vector machines [J]. IEEETransactions on Neural Networks, 2002, 13(2): 464-471.
[41] Suykens J A K, Vandewalle J. Least square support vector machines classifiers[J]. Neural Network Letters, 1999, 19(3):293-300.
[42]张学工.关于统计学习理论与支持矢量机[J].自动化学报,2000 , 26 (1).32-42.
[43]冯瑞,张浩然,邵惠鹤.基于SVM的软测量建模[J].信息与控制,2002,31(6):567-5711.
[44]王定成,方廷健,高理富.支持向量机回归在线建模与应用[J].控制与决策,2003,18(1):89-931.
[45]马勇,黄德先,金以慧.基于支持向量机的软测量建模方法[J].信息与控制,2004,33(4):120-125.
[46]阎威武,朱宏栋,邵惠鹤.基于最小二乘支持向量机的软测量建模[J].系统仿真学报,2003,15(10):1494-1496.
[47]王旭东,邵惠鹤.基于神经网络的通用软测量技术[J].自动化学. 1998, 24(5): 702-706.
[48] Shaw A.M., Doyle III F.J. Multivariable nonlinear control applications for a high purity distillation column using a recurrent dynamic neuron model[J]. J.Proc. Cont. .1997, 7 (4) :255-262.
[49] Joseph B and Brosilow C B. Inferential control of processes: Part I Steady state analysis and design[J]. AIChE J. 1978, 15(2): 150-159.
[50]张磊,胡春,钱锋.BP改进算法及其在乙二醇精制软测量中的应用[J].自动化仪表,2005,26(6):31-34.
[51]杨凌,刘进军,金强.基于改进RBF网络的乙烯纯度软测量建模方法[J].微计算机信息,2008,24(10):158-200.
[52]黄江平,陶慧汇,朱志高.基于BP网络的醋酸乙烯聚合率软测量研究[J].机床与液压,2009,37(8):286-287.
[53]庄利锋,杨慧中,张志强.一种分布式模糊神经网络在熔融指数软测量中的应用[J].化工自动化及仪表,2004,31(4):16-18.
[54] Fikri Akdeniz. The Examination And Analysis of Residuals For Some Biased Estimatorsin Linear Regression[J]. Commun. Statist. Theory Meth , 2001,30(6),1171-1178.
[55]章卫国,杨向忠.模糊控制理论与应用[M].西安:西北工业大学出版社, 1990: 34-58.
[56]刘瑞兰,苏宏业,褚健.基于改进模糊神经网络的软测量建模方法[J].信息与控制, 2003, 32(4): 367-370.
[57]田华阁,车荣杰,王平.基于机制建模与模糊建模相结合的聚丙烯熔融指数软测量方法[J].中国石油大学学报, 2009, 33(3): 162-165.
[58] Kohonen T. Self-Organizing Maps[M]. Berlin: Springer-Verlag, 1997, 50-75.
[59] Kosko B. Neural networks and fuzzy systems[M]. Upper Saddle River :Prentice Hall, Inc, 1992, 32-45.
[60] Mhasker H N. Neural networks for optimal approximation of smooth and analytic functions[J]. Neural Computation, 1996, 32(8): 1731-1742.
[61] Poggio T, Girosi F. Regilarization algorithms for learning that are equivalent to multilayer networks[J]. Science, 1990, 1247(32): 978-982.
[62]史忠植.神经网络[M].北京:高等教育出版社, 2009, 48-63.
[63] Haykin S. Adaptive Filter Theory. 3rd. Upper Saddle River[M]. NJ: Prentice Hall, 1996, 56-89.
[64] Hornik K, Stinchcombe M, White H. Multilayer feedforward networks are universal approximators[J]. Neural Networks, 1989, 2(5): 359-366.
[65] Naguyen D and Widrow B. Improving the learning speed of 2-Layer Neural Networks by Choosing Initial Values of Adaptive Weights[C]. Proceedings of the International Joint Conference On Neural Networks. San Diego, CA,1990(3): 21-26.
[66] Dayan P. Recurrent sampling models for the Helmholtz machine[J]. Neural Computational 1999, 11(9): 653-677.
[67]傅庚,唐小我,曾勇.广义递归方差倒数组合预测方法研究[J].电子科技大学学报, 1995, 24(2): 211-217.
[68]陈友华.基于预测有效度的组合预测模型研究[J].预测, 1997, 16(3): 72-73.
[69]陈友华,侯定丕.基于预测有效度的优性组合预测模型研究[J].中国科技大学学报, 2002, 32(2): 172-180.
[70]王应明.基于相关性的组合预测方法研究[J].预测, 2002, 21(2): 58-62.
[71]王应明,罗英.调和平均组合预测中的参数估计技术[J].系统工程与电子技术, 1997, 19(10): 45-49.
[72]王应明.广义加权几何平均组合预测方法研究[J].厦门大学学报(自然科学版), 1998, 37(1): 6-10.
[73] Wolpert D H. Stacked generalization [J]. Neural Networks, 1992, 5(2): 241-259.
[74] Sridhar D V, Bartlett E B, Seagrave R C. Information theoretic approach for combining neural network process models[J]. Neural Networks, 1999,12(6): 915-926.
[75] Zhang J. Developing robust non-linear models through bootstrap aggregated neural networks [J]. Neurocomputing, 1999, 25(1): 93-113.
[76] Sridhar D V,Seagrave R C, Bartlett E B. Process modelling using stacked neural networks[J ]. AIChE J, 1996, 42(9): 2529-2539.
[77] Hashem S. Optimal linear combinations of neural networks[J]. Neural Networks, 1997, 10(4): 599-614.
[78] Zhang J, Morris A J, Martin E B, et al. Prediction of polymer quality in batch polymerization reactors using robust neural networks[J]. Chem Eng J , 1998, 69(2): 135-143.
[79]钟国平.浅述我国聚丙烯工业发展[J].石油化工, 1996, 25(10): 739-744.
[80]杨维洪.聚丙烯发展近况[J].石油化工, 2001, 30(4): 319-324.
[81]黄葆同,陈伟主编,茂金属催化剂及其烯烃版合物[M].北京:化学工业出版社, 2000, 134-145.
[82]洪定一主编,聚丙烯——原理、工艺和技术[M],北京:中国石化出版社, 2002, 370-377.
[83]吕砚山,赵正琦. BP神经网络的优化及应用研究[J].北京化工大学学报, 2001, 31(1): 67-69.
[84]张德贤.前向神经网络合理隐含层结点个数估计[J].计算机工程与应用, 2005, 30(5): 21-24.
[85] Franco Scarselli, Ah Chung Tsoi. Universal approximation using feed-forward neural networks: A survey of some existing methods and some new results[J].Neural Networks. 1998, 11(1): 15-37.
[86] S V Kamarthi, S Pittner. Accelerating neural network training using weight extrapolation[J]. Neural networks. 1999, 12(12): 1285-1299.