Generating dynamic fuzzy models for prediction problems

In this paper we present a new method to generate interpretable fuzzy systems from training data. A fuzzy system is developed for nonlinear systems modeling and for system state forecasting. The antecedent partition uses triangular sets with 0.5 interpolations avoiding the presence of complex overlapping that happens in other methods. Singleton consequents are employed and least square method is used to adjust the consequents. This approach is not a hybrid system and does not employ other techniques, like neural network or genetic algorithm. Two benchmark problems have been used to illustrate our approach: the first one is an input-output NARMAX model, which is one of the most popular models in the neural and fuzzy literature; the second one is the chaotic, nonperiodic and nonconvergence Mackey-Glass series, commonly used to evaluate a time series forecasting scheme.