An Interval Type-2 Fuzzy Innovative Adaptive Extended Kalman Filter for State Estimation

A methodology for state estimation using an innovative adaptive extended Kalman filter based on interval type-2 Mamdani fuzzy is proposed. Interval type-2 fuzzy theory is considered for the innovative adaptive extended Kalman filter, allowing the filter to adapt to changes in the environment noise. The proposed approach provides a method of updating the measurement noise covariance online to dynamically adjust the extended Kalman filter parameters. Aiming to illustrate the efficiency and applicability of the proposed methodology, we carry out simulations for estimating the states of a two-phase permanent magnet synchronous motor, considering two different conditions: in the first condition, the measurement noise covariance matrix is constant and unknown; in the second, the covariance matrix changes dynamically during the state estimation according to a first-order Markov model, simulating a real-world environment. Numerical results show that the proposed methodology performs well in terms of precision and robustness, even if the noise variance of measurement changes during the dynamical system state estimation.