基于修正闭环子空间辨识--分段线性结构的 环管式丙烯聚合反应过程非线性模型预测控

娄海川; 苏宏业; 古勇; 侯卫锋; 谢磊; 荣冈

引用本文:	娄海川,苏宏业,古勇,侯卫锋,谢磊,荣冈.基于修正闭环子空间辨识--分段线性结构的环管式丙烯聚合反应过程非线性模型预测控[J].控制理论与应用,2015,32(8):1040~1051.[点击复制]
	LOU Hai-chuan,SU Hong-ye,GU Yong,HOU Wei-feng,XIE Lei,RONG Gang.Nonlinear predictive control with modified closed-loop subspace identification-piecewise linear model for double-loop propylene polymerization process[J].Control Theory and Technology,2015,32(8):1040~1051.[点击复制]

基于修正闭环子空间辨识--分段线性结构的环管式丙烯聚合反应过程非线性模型预测控

Nonlinear predictive control with modified closed-loop subspace identification-piecewise linear model for double-loop propylene polymerization process

摘要点击 2521 全文点击 1316 投稿时间：2014-10-14 修订日期：2015-05-29

查看全文查看/发表评论下载PDF阅读器

DOI编号 10.7641/CTA.2015.40902

2015,32(8):1040-1051

中文关键词非线性模型预测控制 MSSARX–PWL结构双环管丙烯聚合反应过程分段线性化

英文关键词 nonlinear model predictive control MSSARX–PWL structure double-loop propylene polymerization process piece wise linear

基金项目

作者	单位	E-mail
娄海川	浙江大学工业控制技术国家重点实验室智能系统与控制研究所浙江中控软件技术有限公司	sdnjchq@zjut.edu.cn
苏宏业	浙江大学工业控制技术国家重点实验室智能系统与控制研究所	shinotang222@gmail.com
古勇	浙江大学工业控制技术国家重点实验室智能系统与控制研究所
侯卫锋	浙江中控软件技术有限公司
谢磊	浙江大学工业控制技术国家重点实验室智能系统与控制研究所
荣冈	浙江大学工业控制技术国家重点实验室智能系统与控制研究所

中文摘要

针对环管式聚丙烯生产过程装置多变量、耦合和非线性等特性容易导致过程控制不稳定及质量指标波动问题, 本文提出了一种基于修正闭环子空间辨识–分段线性(MSSARX--PWL)维纳(Wiener)模型结构的非线性模型预测控制算法. 利用修正的闭环子空间辨识方法(MSSARX)辨识对象在闭环工况下的线性状态空间模型, 并将该线性模型与多变量分段线性化(PWL)方法辨识得到的非线性稳态模型结合, 建立双环管丙烯聚合反应动态过程的非线性预测模型, 而后进一步将非线性模型转化为线性模型, 在线性预测控制算法框架下用二次线性规划方法(LQP)优化控制器, 无须用非线性规划方法(NLP)求解. 从双环管丙烯聚合反应过程仿真例子表明, 该算法不仅能保证模型和控制精度, 而且能提高计算效率.

英文摘要

To solve the unstable and oscillatory problem of the double-loop propylene polymerization process with multivariable, coupling and nonlinearity, we propose a nonlinear model predictive control algorithm based on the Wiener-type model by modified closed-loop subspace identification (MSSARX). A linear state space model under closed-loop conditions and a nonlinear steady-state model are identified by using the modified closed-loop subspace identification method (MSSARX) and the multivariate piecewise linear (PWL) method, respectively. These two models are then combined into an MSSARX–PWL model structure, which is employed as the nonlinear predictive model of the process. To reduce the computational load, this nonlinear model is consequently linearized to ensure that linear quadratic programming (LQP) optimization controller can be applied instead of the nonlinear one. By applying the proposed algorithm, not only the accurate prediction and control are guaranteed, but the computational efficiency is also improved at the same time. The effectiveness of the proposed algorithm is demonstrated on the simulation process of double-loop propylene polymerization.