基于系统辨识与T-S模糊神经网络的磨矿分级控制  

Control of milling classification using system identification and T-S fuzzy neural network

作　　者：赵宏伟[1,2] 齐一名[1] 臧雪柏[1] 张孝临[1] 马英喆[1] 

机构地区：[1]吉林大学计算机科学与技术学院,长春130012 [2]吉林大学符号计算与知识工程教育部重点实验室,长春130012

出　　处：《吉林大学学报（工学版）》2011年第1期171-175,共5页Journal of Jilin University:Engineering and Technology Edition

基　　金：'十一五'国家科技支撑计划项目(2006BAC21B01);鞍钢集团科技攻关计划项目(2007-科A23)

摘　　要：针对磨矿分级过程控制中具有的慢时变、非线性特征,提出了一种基于系统辨识的自适应模糊推理网络模型,并应用于磨矿控制领域。利用模糊聚类法对现有数据样本进行系统辨识,自动获取模糊规则库和相应的初始参数。依据得到的模糊系统构建基于Takage-Suge-no推理模型的自适应模糊神经网络推理系统,获得比传统的模糊神经网络具有更强的自适应性和更快的运算速度。仿真实验结果显示,本文设计的控制模型在磨矿控制过程中具有较好的应用效果。Milling classification control is a slow time delaying and nonlinear process. A control method was presented Based on fuzzy neural network. To automatically acquire the fuzzy rules and the initial parameters of the algorithm, the fuzzy clustering was used in system identification. According to the received fuzzy system, an adaptive-network based fuzzy inference system was built on Takage-Sugone inference model with no professional experience, the capability of which is more adaptive and faster than traditional Fuzzy Neural Network （FNN）. Computer simulation of a milling classification control process was carried out, which validates the proposed control method.

关 键 词：人工智能 模糊神经网络 系统辨识 自适应控制 磨矿分级控制 

分 类 号：TP274[自动化与计算机技术—检测技术与自动化装置]