A comparison of various soft computing techniques in model identification of high complexity systems

Fuzzy model identification techniques are presented in this paper. The paper discusses how fuzzy rules from a pattern set can be extracted without human interference. There are several methods used for rule extraction. Some of these were inspired by biological evolution. Other algorithms have been developed initially for neural networks and can be adapted to fuzzy systems. Fuzzy clustering has also been used for rule extraction. The fuzzy rule interpolation method and hierarchical rule bases are introduced. Combining fuzzy rule interpolation with the use of hierarchically structured fuzzy rule bases leads to the reduction of the fuzzy algorithms' complexity. The paper introduces a technique to automatically construct a hierarchical fuzzy system from training data. Copyright (C) 2003 IFAC.