Algoritmo de Agrupamento Fuzzy Evolutivo Baseado em Critério de Decisão Multivariável para Modelagem Computacional via Dados Experimentais

Abstract

This paper proposes an evolving fuzzy clustering algorithm for time series. It is based on the multivariable Gaussian potential function criterion, which includes the rate of variation as information in this context. The proposed approach creates groups with different shapes in the evolving context, in addition to incorporating a sensitivity factor and a crossover mechanism in the decision-making regarding the structure of the clusters. The approach is validated using computational data from a complex system and experimental data from a thermal system. It’s also used in an evolving fuzzy system to identify the dynamics of the thermal system. The results show that the algorithm generates clusters that accurately represent the behavior of dynamic systems.