带遗忘因子的线性回归性能评估算法及应用  

Performance Assessment Algorithm of Improved Linear Regression with Forgetting Factor( ILR) and Its Application

作　　者：刘学彦[1] 王昕[2] 王振雷[1] 

机构地区：[1]华东理工大学化工过程先进控制和优化技术教育部重点实验室,上海200237 [2]上海交通大学电工与电子技术中心,上海200240

出　　处：《控制工程》2014年第6期867-872,共6页Control Engineering of China

基　　金：国家自然科学基金项目(61134007);十二五国家科技支撑计划项目(2012BAF05B00);上海市科技攻关项目(12dz1125100);上海市"科技创新行动计划"研发平台建设项目(13DZ2295300);上海市自然科学基金项目(14ZR1421800);流程工业综合自动化国家重点实验室开放课题基金(PAL-N201404)

摘　　要：工业过程中的控制系统经常处于性能不佳的状态,若没有得到定期的维护会造成巨大的经济损失。带遗忘因子的线性回归(ILR)算法是一种基于过程数据的性能评估方法,它在线性回归方法的基础上引入遗忘因子,使评估结果更加具有实时性,为控制系统的维护提供了可靠的依据。该算法具有建模简单、计算快捷和适用于时变扰动模型的优点。本文运用该方法对乙烯裂解炉几个重要回路进行了性能评估,以此来表征系统整体的性能,并验证了算法的时变有效性。In the industrial process, control system is often in poor performance status, without regular maintenance, system would suf- fer huge economic losses. Improved linear regression method with forgetting factor （ILR） is a performance evaluation method based on process data. This method makes assessment resuh＇s real-time performance better by bringing in forgetting factor on base of linear re- gression method, providing a reliable basis for the maintenance of the control system. This algorithm has the advantages of establish model simply, calculate quickly and suitable for time-varying disturbance. Finally, it is applied to evaluating the control performance of several important control loops in ethylene cracking furnace, thereby characterizing the overall performance, which verifies the time-var- ying validity of the algorithm.

关 键 词：遗忘因子 线性回归 性能评估 乙烯裂解炉 时变扰动 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]