Methodology for modeling fuzzy Kalman filters of minimum realization from evolving clustering of experimental data.

A methodology for evolving fuzzy Kalman filter identification, is proposed in this paper. The mathematical formulation contemplates the following aspects: for initial estimation, an offline GK clustering algorithm and an offline fuzzy version of OKID algorithm are used to estimate antecedent and consequent parameters, respectively. From each new sample of input-output experimental data from dynamical system, the evolving eTS algorithm and an evolving fuzzy version of OKID algorithm are used to estimate the antecedent and consequent parameters of the evolving fuzzy Kalman filter, respectively. Computational and experimental results considering the estimation of states and outputs of a nonlinear dynamic system and a 2DoF helicopter, respectively, show the efficiency and applicability of the proposed methodology.