基于Fisher判别分析的过程运行状态在线评价  

Online assessment of process operating performance based on Fisher discriminant analysis

作　　者：刘炎[1] 常玉清[1,2] 王福利[1,2] 

机构地区：[1]东北大学信息科学与工程学院,沈阳110004 [2]东北大学流程工业综合自动化国家重点实验室,沈阳110004

出　　处：《控制与决策》2014年第9期1655-1660,共6页Control and Decision

基　　金：国家自然科学基金项目(61374146;61174130);国家863计划项目(2011AA060204);流程工业综合自动化国家重点实验室基础科研业务费项目(2013ZCX0204);中央高校基本科研业务费专项资金项目(N120404020)

摘　　要：为了实时掌握生产过程运行状态,提出一种基于Fisher判别分析(FDA)的过程运行状态在线评价方法.提出离线数据分类与识别算法,以识别不同稳定运行状态的建模数据及其对应的状态等级;利用FDA提取各个稳定运行状态的特征属性,建立评价模型;在线评价时,通过"时间窗口"数据特征与各个状态等级的相似度,实时评价过程运行状态.将所提出的方法应用于某湿法冶金过程的仿真结果验证了该方法的有效性.In order to grasp the process operating performance in real time, an online assessment method for process operating performance based on Fisher discriminant analysis（FDA） is proposed. The modeling data corresponding to different operating performances is separated and identified by the algorithm of offline modeling data classification and identification. The approach extracts the characteristics contained in different operating performances by FDA and establishes the assessment models. In online assessment, the process operating performance can be evaluated by the similarities between the characteristics of sampling window and each performance grade. The proposed method is used to evaluate the process operating performance of a hydrometallurgical process, and the simulation results show its effectiveness.

关 键 词：FISHER判别分析 运行状态 在线评价 湿法冶金 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]