摘要
熔融指数(MI)是聚丙烯生产的重要指标之一,建立可靠的熔融指数预报模型非常重要。丙烯聚合过程机理复杂,生产工艺多样,设备工段繁多,因此采用机理建模方法存在相当大的难度。而统计建模方法是一种依赖数据的方法,对系统内部机理的了解要求很少,所以在丙烯聚合熔融指数预报中得到广泛应用。
统计学习理论是一种基于小样本的机器学习理论,支持向量机是在此理论基础上提出来的。它根据结构风险最小化原则,通过核函数在一个高维特征空间中构造线性决策函数,避免了维数灾难,且可达到全局最优解。支持向量机良好的性能使其成为机器学习领域的热点课题。支持向量机的性能依赖其参数的选择,本文应用智能优化算法进行参数寻优,从而建立多种智能支持向量机模型。本文的主要内容包括:
1.为了降低标准支持向量机(SVM)的计算复杂度,提高其学习速度、泛化能力和稀疏性,本文研究了最小二乘支持向量机(LSSVM).加权最小二乘支持向量机(WLSSVM)和相关向量机(RVM),并用于熔融指数预报。实验结果表明上述方法的可行性和有效性。
2.核函数参数和惩罚参数决定了支持向量机的性能,选择最佳的参数可以直接提高模型预报能力。针对支持向量机的参数选择问题,本文采用带有权重因子的改进粒子群优化算法,分别对LSSVM、WLSSVM和RVM的参数进行寻优,建立了单纯智能支持向量机模型(PSO-LSSVM、PSO-WLSSVM和PSO-RVM)。PSO算法具有很强的寻优能力和快速的收敛速度,能在最短的时间内找到函数的全局最优点,使参数寻优成为可能。实验结果表明优化后的模型具有更好的预报效果。
3.针对标准粒子群算法在迭代过程中易出现粒子过早收敛而陷入局部最优的缺陷,通过引入免疫系统的抗体选择机制,构造了一种基于免疫机制的免疫粒子群优化(IC-PSO)算法,来保持更新粒子的多样性,从而克服标准粒子群算法过早收敛的缺陷;为了减小粒子群搜索的盲目性,避免早熟,本文利用蚁群算法为免疫粒子群算法找到一条最优路径,构造了蚁群-免疫粒子群优化(AC-ICPSO)算法。然后利用这两种优化方法对LSSVM和WLSSVM进行参数寻优,建立了混合智能支持向量机模型(ICPSO-LSSVM、AC-ICPSO-LSSVM. ICPSO-WLSSVM、AC-ICPSO-WLSSVM).以实际聚丙烯生产的熔融指数预报作为实例进行研究,结果表明所提出模型的有效性和良好的预报精度。
4.针对丙烯聚合生产控制中,系统存在高复杂性、不可确定性、多层次性等特点,本文提出了模型在线校正策略。随着数据的更新,不断调整预报模型以适应最新工况。实验表明,校正后的模型预报效果更好。
Melt index (MI) is considered as one of the important variables of the quality in the propylene (PP) polymerization process, and it is crucial to propose a reliable estimation model of MI.. PP polymerization process usually involves complex kinetic mechanism, and various plants, which makes it a challenge to study the process though mechanism modeling approaches. Based on the dada, the statistical method doesn't need much knowledge of the studied process, so it is used to predict MI popularly.
Statistical Learning Theory (SLT) is a theatrical framework of machine learning for small samples, and Support Vector Machine (SVM) comes out from this theory. According to the structural risk minimization rule, it can get the global optimal linear decision function in a higher dimensional feature space according to a kernel function. It avoids the curse of dimensionality and it is of good generalization ability. Its good performance makes it become a hot topic in machine learning. However, its good performance depends on the chosen of parameters, and this paper makes the use of intelligent optimization algorithms to optimize the parameters, so many kinds of intelligent SVM models are proposed. The main contributions of the present work are as follows:
1. In order to improve the training speed、generalization ability and space ability of the traditional SVM, this paper does some research on Least Square SVM (LSSVM)、Weighted LSSVM (WLSSVM)、Relevance Vector Machine (RVM), and these methods are used to predict MI. The results confirm the models'validity.
2. The kernel parameter and the regularization factor determine the performance of SVM, so obtaining the best parameters can improve the prediction ability of models directly. In order to choose better parameter for SVM,this paper applies improved Particle Swarm Optimization (PSO) algorithm, which has an inertia weight factor, to optimize the parameters of LSSVM. WLSSVM and RVM, so the pure intelligent SVM models (PSO-LSSVM. PSO-WLSSVM and PSO-RVM) are proposed. The good search capability and fast convergence make PSO algorithm applicable to parameter optimization. The results show that the optimal models have better prediction ability.
3. Addressing the deficiency of the Particle Swarm Optimization (PSO) algorithm whose particles are easy to sink into premature convergence and run into local optimization in the iterative process, with the clone selection strategy of immune system, the Immune Clone PSO (ICPSO) algorithm based on immune strategy was proposed to make the particles of ICPSO maintain the diversity during the iterative process so as to overcome the defect of premature convergence of PSO; in order to reduce the blindness of the research of PSO and avoid premature, Ant Colony Optimization (ACO) algorithm is used to find an optimal path for ICPSO algorithm, thus the AC-ICPSO algorithm is proposed. Then the two algorithms are used to optimize the LSSVM and WLSSVM's parameters, so the hybrid intelligent SVM models (ICPSO-LSSVM、AC-ICPSO-LSSVM、ICPSO-WLSSVM. AC-ICPSO-WLSSVM) are found. Researches on the optimized model were illustrated with the real plant of propylene polymerization, and the results showed that the proposed approach had great prediction accuracy and validity.
4. In order to avoid the characteristics of high complexity, uncertainty, multi level nature and so on in PP polymerization process, Online Correction Strategy (OCS) is presented in this paper. With the update of the data, the forecast models need adjustment constantly to adapt the latest conditions. The experiments show that the corrected models have better prediction ability.
引文
[1]Mitchell T. Machine Learning [M]. New York:McGraw-Hill Inc,1997
[2]Haykin S., Lustman F. Neural networks and learning machines [M]. New Jersey: Prentice Hall,2008
[3]Mitechell M. T. (曾华军,张银奎译).机器学习[M].北京:机械工业出版社,2003
[4]王珏.机器学习及其应用丛书:关于机器学习的讨论[M].北京:清华大学出版社,2006
[5]De A. Araujo R., Ferreira Tiago A. E. An intelligent hybrid morphological-rank-linear method for financial time series prediction [J]. Neurocomputing,2009,72(10-12):2507-2524
[6]Zhao L. H., Guo Z. K. Face recognition method based on adaptively weighted block-two dimensional principal component analysis[C]. Proceedings of the 3rd International Conference on Computational Intelligence, Communication Systems and Networks, CICSyN,2011:22-25
[7]Jin L., Lin J. L., Lin K. P. Precipitation prediction modeling using neural network and empirical orthogonal function base on numerical weather forecast production[C]. Proceedings of the World Congress on Intelligent Control and Automation (WCICA), 2006:2723-2727
[8]Ince T., Kiranyaz S., Gabbouj M. Evolutionary RBF classifier for polarimetric SAR images[J]. Expert Systems with Applications,2012,39(5):4710-4717
[9]Arcia A. P., Oliver J., Gosch, D. An intelligent agent-based distributed architecture for Smart-Grid integrated network management[C]. Proceedings of Conference on Local Computer Networks, LCN,2010:1013-1018
[10]Vapnik V. N., Chervonenkis A. Y. On the uniform convergence of relative frequencies of events to their probabilities [J]. Theory of Probability and its Applications,1971,16(2):264-280
[11]Vapnik V. N., Lemer A. Pattern recognition using generalized portrait method [M]. Automation and Remote Control,1963,24
[12]Vapnik V. N. Statistical Learning Theory [M]. New York:Wiley,1998.
[13]Gao J. B., Gunn S. R., Harris C. J. SVM regression through variational methods and its sequential implementation[J]. Neurocomputing,2003,55(1-2):151-167
[14]李建.工程结构优化的群体智能算法[D].硕士学位论文,浙江大学,2010
[15]Huang W., Xu C. L., Zhang J. H. et al. Study of reactive power optimization based on immune genetic algorithm[C]. IEEE PES Transmission and Distribution Conference and Exposition,2003,1:186-190
[16]曹劲,王桂增,徐博文.基于鲁棒自适应RBF网络的聚丙烯熔融指数预报[J].控制与决策,1999,14(4):339-343
[17]梁林,李春富,王桂增.非线性递推部分最小二乘及其应用[J].系统仿真学报,2001,13(Suppl.):119-121,125
[18]李春富,王桂增,徐博文.聚丙烯熔融指数软测量[J].化工自动化及仪表,2002,29(5):22-25
[19]李春富,王桂增,叶昊.基于操作域划分的聚丙烯熔融指数软测量[J].化工学报,2005,56(10):1915-1921
[20]孔薇,杨杰.基于神经网络的非线性PCA方法[J].计算机仿真,2003,20(7):65-67
[21]孔薇,杨杰.径向基神经网络的聚丙烯熔融指数预报[J].化工学报,2003,54(8):1160-1163
[22]Han I. S., Han C., Chung C. B. Melt index modeling with support vector machines, partial least squares, and artificial neural networks [J]. Journal of Applied Polymer Science,2005,95(4):967-974
[23]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, 70:280-287
[24]Shi J., Liu X. G. Melt index prediction by neural soft-sensor based on multi-scale analysis and principal component analysis [J]. Chinese Journal Chemical Engineering,2005,13(6):849-852
[25]Shi J., Liu X. G. Melt index prediction by weighted least squares support vector machines [J]. Journal of Applied Polymer Science,2006,101(1):285-289
[26]楼巍,刘兴高.基于PCA-GA-RBF网络的聚丙烯熔融指数预报模型[J].石油化工高等学校学报,2007,20(3):82-85
[27]Zhao C. Y., Liu X. G., Ding F. Melt index prediction based on adaptive particle swarm optimization algorithm-optimized radial basis function neural networks[J]. Chemical Engineering and Technology,2010,33(11):1909-1916
[28]李九宝,刘兴高.基于PSO_SA算法的聚丙烯熔融指数预报[J].化工学报,2010,61(8):1955-1959
[29]Jin X. L., Hu P. H., Yuan P., Dong S. P. Development of PP melt index inferential model [C]. Proceedings of the World Congress on Intelligent Control and Automation (WCICA),2006,1:4866-4870
[30]庄利锋,杨慧中等.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
[31]Wang S. M., Xu J. Y., Zhai Y. Q. Generalized predictive control of neural network based on LM optimization[J]. Applied Mechanics and Materials,2011, 66-68:2164-2169
[32]Kang M. G., Park S. W., Cai X. M. Integration of hydrologic gray model with global search method for real-time flood forecasting[J]. Journal of Hydrologic Engineering,2009,14(10):1136-1145
[33]Vapnik V. N. An overview of statistical learning theory[J]. IEEE Transactions on Neural Networks,1999,10(5):988-999
[34]包哲静,皮道映,孙优贤.基于并行支持向量机的多变量非线性模型预测控制[J].控制与决策,2007,22(8):922-926
[35]张日东,王树青,李平.基于支持向量机的非线性系统预测控制[J].自动化学报,2007,33(10):1066-1073
[36]李元乐,陶兰.基于小波核支持向量机的蛋白质二级结构预测[J].深圳大学学报理工版,2006,23(2):117-121
[37]郑永康,陈维荣,戴朝华.小波支持向量机与相空间重构结合的短期负荷预测研究[J].继电器,2008,36(7):29-33
[38]Suykens J. A. K., Vandewalle J. Least squares support vector machine classifiers[J]. Neural Processing Letters,1999,9:293-300
[39]Suykens J. A. K., Vandewalle J. Weighted least squares support vector machines: robustness and sparse approximation [J]. Neurocomputing,2002,48:85-105
[40]Tipping M. E. The relevance vector machine[C]. Advances in Neural Information Processing Systems, Cambridge,2000:652-658
[41]Tipping M. E. Sparse Bayesian learning and the relevance vector machine [J]. Journal of Machine Learning Research,2001,1:211-244.
[42]MacKay D. J. C. The evidence framework applied to classification networks [J]. Neural Computation,1992,4(5):720-736
[43]Kennedy J., Eberhart R. Particle swarm optimization[C]. Proceedings of IEEE International Conference on Neural Networks,1995,4:1942-1948
[44]Bafna, S.S., Beall, A.-M. Design of experiments study on the factors affecting variability in the melt Index measurement[J]. Journal of Applied Polymer Science, 1997,65:277-288
[45]Hinchliffe M., Montague G., Willis M., Burke, A. Hybrid Approach to Modeling an Industrial Polyethylene Process [J]. AIChE Jounal,2003,49(12):3127-3137
[46]蒋京波,韦建利,徐用懋.聚丙烯熔融指数的在线计算及工程应用[J].计算机工程与应用2002,18:220-223
[47]Wise B. M., Gallagher N. B. The process chemometrics approach to process monitoring and fault detection [J]. Journal of Process Control,1996,6(6):329-348
[48]焦李成,刘芳等.智能数据挖掘与知识发现[M].西安电子科技大学出版社,2006
[49]Eberhart R. C., Kennedy J. A new optimizer using particles swarm theory[C]. Proceedings of the Sixth International Symposium on Micro Machine and Human and Human Science,1995:39-43
[50]Bermolen P., Rossi D. Support vector regression for link load prediction [J]. Computer Networks,2009,53(2):191-201
[51]Chappelle O., Vapnik V. N., Bousquet O., et al. Choosing multiple parameters for support vector machines [J]. Machine Learning,2002,46(1):131-160
[52]Chen P. w., Wang J. Y., Lee H. Model selection of SVMs using GA approach [C]. Proceedings of 2004 IEEE Int Joint Confon Neural Networks. Piscataway, NJ:IEEE Press,2004:2035-2040
[53]Zheng C. H., Jiao L. C. Automatic parameters selection for SVM based on GA[C]. Proceedings of the 5th World Congress on Intelligent Control and Automation. Piscataway, NJ:IEEE Press,2004:1869-1872
[54]Wang J., Gao X. G., Zhu Y. W. Solving algorithm for TA optimization model based on ACO-SA[J]. Journal of Systems Engineering and Electronics,2011,22(4): 628-639
[55]Liu B., Wang L., Jin Y. H. et al. Improve particle swarm optimization combined with chaos [J]. Chaos, Solitons and Fractals,2005,25:1261-1271
[56]Yi D., Ge X. R. An improved PSO-based ANN with simulated annealing technique [J]. Neurocomputing,2005,63:527-533
8[57]修春波,张雨虹.基于蚁群与鱼群的混合优化算法[J].计算机工程,2008,34(14):206-217
[58]龚非力.医学免疫学[M].北京:科学出版社,2003,1:1-27
[59]D'haeseleer P., Forrest S., Helman P. An immunological approach to change detection algorithms:Analysis and implication[C]. Proceedings of IEEE Symposium on Security and Privacy, Las Alamitos, CA, USA,1996:110-119
[60]De Castro L. N., Von Zuben F. J. Learning and optimization using the clonal selection principle[J]. IEEE Transactions on Evolutionary Computation,2002,6(3): 239-251
[61]席裕庚,恽为民.遗传算法综述[J].控制理论与应用,1996,13(6):697-708
[62]葛红,毛宗源.免疫算法的改进[J].计算机工程与应用,2002,(14):47-49
[63]Colorni A., Dorigo M., Maniezzo V. Distributed optimization by ant colonies[C]. Proceedings of the First European Conference on Artificial Life,1992: 134-142
[64]Dorigo M., Maniezzo V., Colorni A. Optimization by a colony of cooperating Agents[J]. IEEE Transactions on Systems, Man and Cybernetics, Part B,1996, 26(1):29-41
[65]Dorigo M., Gambardella L. M. A cooperative learning approach to the travelling salesman problem [J]. IEEE Transactions on Evolutionary Computation, 1997,1(1):53-66
[66]Schoonderwoerd R., Bruten J. L., Holland O. E. Ant-based load balancing in telecommunications networks [J]. Adaptive Behavior,1996,5(2):169-207
[67]Gianni D. C., Dorigo M. Distributed stigmergetic control for communications Network [J]. Journal of Artficial Intelligence Research,1998,9:317-365
[68]李红军,秦永胜,徐用懋.化工过程中的数据协调及显著误差检测[J].化工自动化及仪表,1997,24(2):25-32
[69]王秀萍,荣岗,王树青.先进控制技术及应用[J].化工自动化及仪表,1999,26(3):62-67
[70]Kuehn D. R., Davidson H. Computer control II, mathematics of control [J]. Chemical Engineering Progress,1961,57(6):44-47
[71]Narasimhan S., Harikumar P. A method to incorporate bounds in data reconciliation and gross error detection-Ⅰ. The bounded data reconciliation problem [J]. Computers and Chemical Engineering,1993,17(11):1115-1120
[72]Harikumar P., Narasimhan S. A method to incorporate bounds in data reconciliation and gross error detection II. Gross error detection strategies [J]. Computers and Chemical Engineering,1993,17(11):1121-1128
[73]Dovi V. G., Reverberi A. P., L. Maga L. Reconciliation of process measurements when data are subject to detection limits [J]. Chemical Engineering Science,1997, 52(17):3047-3050
[74]Keller J. Y., Zasadzinski M., Darouach M. Analytical estimator of measurement variances in reconciliation [J]. Computers and Chemical Engineering,1992,16(3): 185-188
[75]Chen J., Bandoni A., Romagnoli A. Robust estimation of measurement error variance/covariance from process sampling data[J]. Computers and Chemical Engineering,1997,21(6):593-600.
[76]Knepper J. C., Gorman J. W. Statistical analysis of constrained data sets [J]. AIChE Journal,1980,26:260-264
[77]Crowe C. M. Reconciliation of process flow rates by matrix projection, part Ⅱ, the nonlinear case [J]. AIChE Journal,1986,32(4):616-623
[78]Stanley G. M., Mah R. S. H. Estimation of flows and temperatures in process network [J]. AIChE Journal,1977,23(5):642-650
[79]Chen J., Bandoni A., Romagnoli J. A. Outlier detection in process plant data [J]. Computers and Chemical Engineering,1998,22(4/5):641-646
[80]Ramamurthi Y., Sistu R. B., Bequette B. W. Control-relevant dynamic data reconciliation and parameter estimation[J]. Computers and Chemical Engineering, 1993,17(1):41-59
[81]Tona R.V., Benqlilou C., Espuna A., Puigjaner L., Dynamic data reconciliation based on wavelet trend analysis [J]. Industrial and Engineering Chemistry Research, 2005,44,4323-4335
[82]李红军,秦永胜,徐用懋.数据校正在精馏塔系中的应用[J].化工自动化 及仪表,1997,24(1):30-36
[83]Kong M., Chen B., Li B. An integral approach to dynamic data rectification [J]. Chemical Engineering Science,2000,24:749-753
[84]Zhang P., Rong G. Steady-state bilinear data reconciliation dealing with scheduling [C]. Proceedings of 15th IFAC World Congress,2002
[85]Crowe, C. M. Data reconciliation-progress and challenges[J]. Journal of process control,1996,6(2-3):89-98
[86]Reilly P. M., Carpani R. E. Application of statistical Theory to adjustment of material balances [C]. Proceedings of the 13th Can. Chem. Eng. Conf., Montreal, Quebec,1963
[87]Almasy G. A., Sztano T. Checking and correction of measurements on the basis of linear system model [J]. Problems of Control and Information Theory,1975,4(1): 57-69
[88]Mah R. S. H., Stanley G. M., Downing D. W. Reconciliation and rectification of process flow and inventory data [J]. Industrial and Engineering Chemistry Process Design and Development,1976,15:175-183
[89]Tong H., Crowe C. M. Detecting persistent gross errors by sequential analysis of principal components [J]. AIChE Journal,1996,43(5):1242-1249
[90]王希若,荣冈.测量仪表过失误差的单结点识别方法[J].化工学报,2000,51(1):17-22
[91]王希若,荣冈.高置信度的显著误差综合检测法[J].仪器仪表学报,2000,21(2):196-200
[92]Zhang P., Rong G., Wang Y. A new method of redundancy analysis in data reconciliation and its application [J]. Computers and Chemical Engineering,2001, 25(7-8):941-949
[93]Chen J., Bandoni A., Romagnoli J. A. Outlier detection in process plant data [J]. Computers and Chemical Engineering,1998,22(4/5):641-646
[94]Kong M., Chen B., He X., Hu S. Gross error identification for dynamic system [J]. Computers and Chemical Engineering,2004,29:191-197
[95]卢荣德,陈宗海,王雷.思维模拟在复杂工业过程建模中应用的综述[J].信息和控制,2002,31(4):346-352
[96]钱峻,邵惠鹤.一种小波神经网络的在线建模和校正算法[J].模式识别与人工智能,2000,13(1):16-20
[97]陶少辉,史书阳,刘猛,金思毅LS-SVM模型在线校正的替代法及其软测量应用[J].2010,37(8):15-18
[98]杨尔辅,周强,胡益峰,徐用懋.基于PCA-RBF神经网络的工业裂解炉收率在线预测软测量方法[J].系统仿真学报,2001,13(8):194-197
[99]De Kruif B. J., De Vries Theo J. A. Pruning error minimization in least square support vector machines [J]. IEEE Transactions on Neural Networks,2003,14(3): 696-702
[100]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):524-532
[101]陶少辉,史书阳,刘猛,金思毅LS-SVM模型在线校正的替代法及其软测量应用[J].化工自动化及仪表,2010,37(8):15-18