一种基于改进设计性能指标的模型预测控制器性能监控方法  

Performance assessment of model predictive controller by design-case benchmark based on statistical process control

作　　者：刘伟[1] 田学民[1] 

机构地区：[1]中国石油大学(华东)信息与控制工程学院,山东东营257061

出　　处：《计算机与应用化学》2011年第7期858-862,共5页Computers and Applied Chemistry

基　　金：国家863计划项目(编号:2007AA04Z193);山东省自然科学基金资助项目(编号:Y2007G49)

摘　　要：受随机噪声的影响,传统的设计性能指标是一个自相关的随机变量,传统的统计过程控制方法无法对其进行有效的监控。针对此问题,本文提出一种基于改进设计性能指标的模型预测控制器性能监控方法。为了对控制系统一段时间内的整体性能做出有效评价,首先将性能指标转化为平均形式,建立该性能指标的时间序列模型并对其进行预测,对预测误差进行统计过程控制,包括对预测残差及其累积和的监控,实现了对控制系统总体性能的实时评价。当检测出控制系统性能出现下降时,通过基于距离的判别分析方法快速定位性能下降源。在Wood-berry塔上的仿真研究验证了方法的可行性与有效性。Because of the stochastic noise, the traditional design-case index is stochastic with autocorrelation. The traditional statistical process control method couldn＇t monitor it effectively. In order to overcome these shortcomings, an improvd design-case index is presented. For assess the general performance of the control system, change the performance index to the average value of a period. Get the time series model of the index, handle the residual with statistical process control to assess the system, including monitoring the predictive error of the index and its cusum chart, to assess the control performance. Diagnose the fault source using the discriminant analysis based on distance. The simulation results on the Wood-Berry model validate the feasibility and effectiveness of proposed method. We set three conditions of the control system, normal, model mismath and changing, the disturbance, the method could detect the deteriorate and find out the reason efficiently.

关 键 词：模型预测控制 性能监控 设计性能基准 统计过程控制 判别分析 

分 类 号：TP277[自动化与计算机技术—检测技术与自动化装置]