摘要
针对目前聚乙烯醇生产过程中醋酸乙烯(VAC)聚合率难以用常规的传感器在线测量的情况,本文在分析了量子粒子群(QPSO)和最小二乘支持回归机(LS-SVM)原理的基础上,提出了利用QPSO和LS-SVM相结合的方法建立醋酸乙烯聚合率软测量模型,利用量子粒子群的全局搜索能力来对最小二乘支持向量机在建模过程中重要的参数进行优化。仿真结果表明:所建立的软测量模型泛化能力强,精度高,比已有的神经网络和支持向量机软测量模型能更好的实现醋酸乙烯聚合率的在线估计。
Aiming at the difficulty in measuring the vinyl acetate ( VAC ) polymerization rate on-line with conventional sensors in the process of polyvinyl alcohol production, the QPSO-LSSVM model about soft sensing of vinyl acetate polymerization rate is established based on the analysis of the quantum particle swarm optimization ( QPSO ) and least squares support regression machine ( LS-SVM ) in this paper, the best parameters of LS-SVM and the mixed kernel function may be selected by QPSO which has the better search ability. The simulation result indicated that the method can obtain a better forecast effect and generalization ability. Compared with the methods base on neural network and support vector machine, the QPSO-LSSVM model is more effective to realize on-line prediction of the vinyl acetate polymerization rate.
出处
《自动化技术与应用》
2011年第8期10-13,17,共5页
Techniques of Automation and Applications
关键词
软测量
醋酸乙烯聚合率
最小二乘支持向量机
量子粒子群优化
soft-sensor
VAC polymerization rate
least squares support vector machine
quantum particle swarm optimization