T-S模型的窑炉压力无超调快速广义预测控制设计  

None-overshoot fast generalized predictive control design for furnace pressure based on TS model

作　　者：雷世昌 

机构地区：[1]北京化工大学信息科学与技术学院

出　　处：《计算机与应用化学》2014年第8期977-981,共5页Computers and Applied Chemistry

摘　　要：窑压是玻璃窑炉运行过程中重要的被控指标之一,直接影响窑炉能耗、寿命及产品成品率,优化窑炉压力控制具有重要的经济意义。由于受到众多因素的影响,窑压具有典型的非线性特性,现有方法的控制效果还有很大的提升空间。本文针对窑压设计了一套新型无超调快速模糊广义预测控制方法(NFGPC),先利用"离线+在线"组合辨识方法得到窑压对象的T-S模型;然后基于该模型对系统进行分片线性化得到时变CARIMA模型,以设计窑压广义预测控制律;结合柔化理论与新型滚动优化目标函数设计一种广义预测控制律,该方法无需求解逆矩阵即可得到控制输出,计算量更小;由于新目标函数的应用,该方法还能够克服传统GPC引起的超调效应。仿真结果表明该方法能够很好的处理窑压非线性系统建模问题,与PID控制、线性GPC(LGPC)以及模糊广义预测控制(FGPC)以及快速模糊广义预测控制(FFGPC)等方法控制效果的对比表明,NFFGPC在处理非线性系统控制问题上具有一定的优越性。Glass furnace pressure is an important indicator of glass production process, which can directly affect furnace＇ energy consumption, longevity and product yield, furnace pressure control optimizing has important economic significance. Furnace pressure has a complex nonlinear characteristics due to many factors, performance of existing control methods still leaves much room for improvement. This paper designed a novel nonlinear None-overshoot Fast Generalized Predictive Control（NFGPC） method. First, a fuzzy ＂offline ＋ online＂ identification strategy was used to get the pressure＇ T-S model; then time variant CARIMA model was obtained by the linearization at each sample point. A novel GPC law was designed based on the combination of soften rolling and new optimization objective function,which has a good property on Overshoot suppression with less computation. Simulation results show that T-S model can be a good modeling method for furnce pressure nonlinear systems. Compared with PID, linear Generalized Predictive Control（LGPC）, Fuzzy Generalized Predictive Control（FGPC）, Fast Fuzzy Generalized Predictive ControI（FFGPC）, NFGPC provide a better control performance.

关 键 词：马蹄焰窑炉 窑压 T-S模型 快速广义预测控制 超调抑制 

分 类 号：TQ015.9[化学工程] TP391.9[自动化与计算机技术—计算机应用技术]