A simple method for identification of singleton fuzzy models

This article presents a simple method for constructing a singleton fuzzy model from a given set of input output data. The tnethod consists of three computational steps: the initial phase, the growth phase, and the optional refining phase. The universe of discourse and two linguistic terms for each input variable and a rule base arc established during the initial phase. Additional linguistic terms and rules are then appended sequentially during the growth phase to modify the model structure and to elevate the performance. During the optional refming phase the overall modelling performance can be further improved by adjusting the singleton outputs of the rule set in the sense of least squares. The proposed identification method can simultaneously provide an appropriate model structure and parameters without any time-consuming optimisation. Several numerical examples demonstrate the effectiveness of the proposed identification method,