基于Smith预估的FNNC-PI复合控制器研究  

Research on the FNNC-PI Compound Controller Based on Smith Prediction

作　　者：薛薇[1] 高建强[1] 贾红艳[1] 

机构地区：[1]天津科技大学电子信息与自动化学院,天津300222

出　　处：《自动化仪表》2006年第10期30-32,共3页Process Automation Instrumentation

基　　金：天津市高等学校科技发展基金项目(编号:20030514)

摘　　要：针对工业过程中常见的大纯滞后现象,提出了一种由模糊神经网络控制(FNNC)与传统PI相结合的复合控制器(FNNC-PI)。由该控制器构成的改进Sm ith预估补偿控制器(Sm ith-FNNC-PI),将模糊控制具有的较强逻辑推理功能、神经网络具有的较强自学习能力以及传统PI控制的优点融为一体。仿真结果表明:提出的控制策略对模型匹配和失配都具有较强的鲁棒性、稳定性,并在动静态特性及抗干扰性等方面具有很好的控制品质。In accordance with phenomena of the large dead time commonly existing in industrial processes, the controller combined with FNNC （ Fuzzy neural networks control ） and the traditional （ PI ） Proportional-Integral is put forward. An improved Smith predictor （ Smith - FNNC - PI ） based on this controller integrates the merits of the logic function of fuzzy control, the serf-learning function of neural network and the advantages of traditional PI control. Results of simulation show the strong robustness and stability of the proposed method in the case of model match and model unmatch. The strategy also offers excellent control performance in static-dynamic state characteristics and anti-interference, etc.

关 键 词：SMITH预估补偿 模糊神经网络控制 大纯滞后 

分 类 号：TP273.3[自动化与计算机技术—检测技术与自动化装置]