基于模式的建模与自适应控制  

Modeling and Adaptive Control Based on Pattern Recognition

作　　者：吴金霞[1] 刘闯[2] 

机构地区：[1]辽宁工业大学理学院,辽宁锦州121001 [2]渤海大学工学院,辽宁锦州121001

出　　处：《数学的实践与认识》2016年第9期203-210,共8页Mathematics in Practice and Theory

基　　金：国家自然科学基金面上项目(11572146);辽宁工业大学校基金(X201321)

摘　　要：针对一类复杂生产过程,提出了一种利用模式的运动来描述系统动力学特性与控制的新方法.方法不同于传统的如输出方程或状态空间模型等建模方法.首先根据实际工况数据利用聚类算法来获得工况模式类别的划分,以这些类别作为空间刻度构造＂模式运动空间＂.然后在此空间上定义描述模式运动的变量—模式类别变量,并建立了基于模式类别变量的预测模型及控制模型,给出了基于模式类别变量的自适应控制方法.最后根据安阳钢铁厂400烧结机的实际工况数据进行了仿真,取得了满意的结果,验证了该种方法的可行性.For a class of complex production processes,a new dynamics description and control method is proposed by using the moving of pattern.This method is different from the traditional modeling methods such as output equation or state space model.First,based on the actual run status data,working condition pattern classes are divided by clustering algorithm.All these classes are used as the space scale to construct ＂pattern moving space＂.Then a variable which we called pattern class variable is defined to describe the moving of pattern.And the prediction model and control model is constructed based on pattern class variable.The adaptive control way based on pattern class variable is also studied.At last,simulation is given based on the actual run status data collected from 400 sintering process of Anyang iron and steel plant.Satisfactory result is obtained and the feasibility is verified.

关 键 词：模式运动 模式运动空间 模式类别变量 自适应控制 

分 类 号：TF046.4[冶金工程—冶金物理化学] TP273[自动化与计算机技术—检测技术与自动化装置]