基于神经网络辨识的模糊预测函数控制  

Predictive function control based on neural networks identification of fuzzy compensation

作　　者：周祥龙[1] 赵景波[2] 

机构地区：[1]青岛海军潜艇学院,山东青岛266071 [2]青岛理工大学,山东青岛266071

出　　处：《电气传动自动化》2010年第2期18-21,共4页Electric Drive Automation

摘　　要：针对生产过程中存在的滞后性、时变性、不确定性和变工况等特点及预测函数控制中模型失配的影响的情况,提出了基于神经网络辨识参数、通过模糊推理对控制量进行补偿的解决方案。并将基于神经网络辨识的模糊补偿预测函数控制应用于锅炉燃烧控制系统,通过连续系统仿真,结果表明这种控制器具有较强的鲁棒性。Considering the characteristics of time delay, time variables, uncertainties and variable operation condi- tion in process control system, solution as compensating the controlled quantity by fuzzy inference is put forward to the problem of model mismatch caused by the predictive function control. The Method Identification based on neural network parameters is also given. Then the predictive-function control of this kind on the basis of neural network identification of fuzzy compensation is applied to the combustion systems of industrial boiler. The simulation of continuous system shows this controller with strong robustness.

关 键 词：神经网络辨识 模糊推理 预测函数控制 锅炉控制 连续系统仿真 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]