Intelligent adaptation of Kalman filters using fuzzy logic

Significant benefits are to be found by dynamically adapting a Kalman filter state estimator if the noise conditions under which it operates change. It is traditional in adaptation schemes to adapt diagonal elements of the process noise covariance matrix, Q(n), or the measurement noise covariance matrix, R(n), or both. A novel adaptive scheme employing the principles of fuzzy expert systems is explored in this paper. The performance of the new scheme is compared with that of two traditional schemes.<>