Title :
Process control using a CFCM-based on-line adaptive neuro-fuzzy systems
Author :
Kwak, Keun-Chang ; Lee, Dae-Jong ; Kim, Sung-Soo ; Ryu, Jeong-Woong
Author_Institution :
Dept. of Electr. Eng., Chungbuk Nat. Univ., South Korea
Abstract :
A conditional fuzzy c-means (CFCM)-based fuzzy adaptive neuro-fuzzy system (ANFS) by on-line learning is proposed in this paper. In the structure identification, the optimal or near optimal number of fuzzy rules is determined by a CFCM clustering with TSK-type fuzzy rules based on the criterion. In the parameter identification. The consequent parameters are tuned by least squares estimator (LSE) and the premise parameters are tuned by back-propagation algorithm in off-line learning. Then on-line learning by recursive least squares estimator (RLSE) and back-propagation algorithm is used to cope with time varying plant dynamics. Finally, we show its capability for a CFCM-based on-line ANFS to control the temperature of a water path
Keywords :
adaptive control; backpropagation; fuzzy control; fuzzy neural nets; least squares approximations; neurocontrollers; parameter estimation; process control; recursive estimation; temperature control; CFCM clustering; TSK-type fuzzy rules; back-propagation algorithm; conditional fuzzy c-means adaptive neuro-fuzzy system; fuzzy rules; least squares estimator; off-line learning; on-line adaptive neuro-fuzzy systems; on-line learning; parameter identification; parameters tuning; process control; recursive least squares estimator; structure identification; time varying plant dynamics; water path temperature control; Adaptive control; Adaptive systems; Clustering algorithms; Fuzzy neural networks; Fuzzy systems; Least squares approximation; Parameter estimation; Process control; Programmable control; Recursive estimation;
Conference_Titel :
Industrial Electronics, 2001. Proceedings. ISIE 2001. IEEE International Symposium on
Conference_Location :
Pusan
Print_ISBN :
0-7803-7090-2
DOI :
10.1109/ISIE.2001.931657
