EFFICIENT FUZZY IDENTIFICATION BASED ON INFERENCE ERROR

Present an efficient methodology to obtain linguistically interpretable fuzzy models using the inference error method. It includes the algorithms to determine classes, rules generation, and partition sum-1 of input variables: shape, number and distribution of fuzzy sets. The load centre of each class represents the output variable’s singleton location. The novelty of our proposal lies on the equilibrium between precision and interpretability, and its low computational complexity. The algorithm was applied to some benchmark classics like the Box-Jenkins’ gas furnace, the Mackey-Glass series and others, getting better results that achieved by others, comparing the MSE, parameter number and interpretability