Método de identificação de transientes com abordagem possibilística, otimizado por algoritmo genético

Abstract

This work develops a method for transient identification based on a possibilistic approach, optimized by Genetic Algorithm to optimize the number of the centroids of the classes that represent the transients. The basic idea of the proposed method is to optimize the partition of the search space, generating subsets in the classes, within a partition, defined as subclasses, whose centroids are able to distinguish the classes with the maximum correct classifications. The interpretation of the subclasses as fuzzy sets and the possibilistic approach provided a heuristic to establish influence zones of the centroids, allowing to achieve the "don't know" answer for unknown transients, that is, outside the training set.