SBR过程自适应动态非线性MPCA建模及在线监视  

Adaptive Dynamic Nonlinear MPCA Modeling and Online Monitoring for Sequence Batch Reactor Process

作　　者：赵立杰[1] 柴天佑[1] 袁德成[2] 

机构地区：[1]东北大学自动化研究中心,辽宁沈阳110004 [2]沈阳化工学院信息工程学院,辽宁沈阳110142

出　　处：《系统仿真学报》2005年第9期2060-2064,2069,共6页Journal of System Simulation

基　　金：国家"863"资助项目(2004AA412030);国家自然科学基金资助项目(60074019)

摘　　要：污水处理SBR过程的在线性能监视对于提高排污质量和实现优化操作是至关重要的。针对SBR间歇过程缓慢时变、可变运行长度和非线性特点,采用双滑动窗口机制,提出多变量非线性自适应动态建模方法—可变长度的滑动窗口MPCA方法,解决常规MPCA在工业应用过程中存在的几个潜在问题:(1)建模数据样本不同步问题;(2)模型失配问题;(3)MPCA线性方法不能充分有效压缩和抽取非线性过程信息;(4)估计未来测量变量所引进的监视误差。所提方法成功应用在国际水协会提出的benchmark仿真试验平台。Online performance monitoring for wastewater treatment SBR batch process is extremely important to improve effluent quality and realize optimal operation. Due to slow varying, variable run length and nonlinear for SBR process, a double moving window MPCA algorithm was proposed to develop an adaptive nonlinear dynamic model for monitoring the progress of SBR processes in real-time. The proposed method easily handles the following problems： （l） Batch durations are unsynchronized; （2） SBR process and its model do not match; （3） Linear MPCA may not be efficient in compressing and extracting nonlinear process data; （4） Errors are added by estimating the future trajectory of the ongoing batch. A benchmark simulation experiment platform for wastewater treatment SBR process published by International Association of Water （IAW） is used to demonstrate the proposed method.

关 键 词：序批式反应器(SBR) 多变量统计过程控制 在线性能监视 滑动窗口MPCA 

分 类 号：TP272[自动化与计算机技术—检测技术与自动化装置]