Filtragem de Kalman Fuzzy Tipo-2 Intervalar Evolutiva Baseada em Processamento Espectral de Dados Experimentais

Abstract

In this doctoral thesis, an interval type-2 evolving fuzzy Kalman filter is proposed for interval processing of uncertain experimental data from unobservable spectral components. The adopted methodology considers the following steps: an initial model of the interval type2 evolving fuzzy Kalman filter is identified offline from an initial window of experimental data; the updating of the antecedent proposition of interval type-2 evolving fuzzy Kalman filter is performed using a interval type-2 fuzzy formulation of the evolving Takagi-Sugeno (eTS) clustering algorithm and the consequent proposition update is performed using a interval type-2 fuzzy formulation of the Observer/Kalman Filter Identification (OKID) algorithm, taking into account the unobservable spectral components extracted from the experimental data by means of an algorithm of Multivariate Recursive Singular Spectral Analysis. The computational results for interval tracking of the Mackey-Glass chaotic time series and for forecasting of the trajectory described by aerospace vehicles, in a noisy environment, illustrate the efficiency of the proposed methodology compared to relevant approaches in the literature. The experimental results regarding the interval tracking of the dynamic behavior of COVID-19 and for filtering and tracking a 2DoF helicopter are presented, in order to elucidate the applicability of the proposed methodology.