Title: An iterative fuzzy identification method hybridising modified objective cluster analysis with genetic algorithm

Authors: Na Wang, Yu-Pu Yang

Addresses: Department of Automation, Shanghai Jiao Tong University, 800 Dongchuan Road, Minhang, Shanghai, 200240, China. ' Department of Automation, Shanghai Jiao Tong University, 800 Dongchuan Road, Minhang, Shanghai, 200240, China

Abstract: In this paper, an iterative fuzzy identification method hybridising modified objective cluster analysis with genetic algorithm (IFI-MOCA&GA) for Takagi-Sugeno-Kang (TSK) fuzzy modelling is proposed. By means of combining the presented modified objective cluster analysis with the fuzzy c-means (FCM) algorithm, the robust and compact fuzzy partition in the input space is obtained. Furthermore, this fuzzy partition is iteratively identified by means of hybridising the modified objective cluster analysis algorithm with the genetic algorithm. Therefore, the accuracy of the fuzzy partition is improved. Using this structure identification strategy, the trade-off between robustness, complexity and accuracy of the model is achieved. Finally, the consequent parameters are estimated by the stable Kalman filter (SKF) algorithm. The performance of the proposed method is shown to be superior to the other methods by a famous electrical simulation example.

Keywords: TSK fuzzy model; fuzzy modelling; structure identification; objective cluster analysis; genetic algorithms; iterative fuzzy identification; fuzzy clustering; Takagi-Sugeno-Kang.

DOI: 10.1504/IJMIC.2010.033843

International Journal of Modelling, Identification and Control, 2010 Vol.10 No.1/2, pp.44 - 49

Published online: 02 Jul 2010 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article