Abstract In recent years, the Kalman filter based on the minimum error entropy (MEE) criterion has been proposed, which outperforms the traditional Kalman filter in the presence of non-Gaussian noise. In practical applications, the estimated performance of the MEE unscented Kalman filter (MEE-UKF) algorithm is influenced by the kernel bandwidth (KB). In addition, it may be unstable in numerical computation. This paper proposes an adaptive robust MEE unscented Kalman filter (AMEE-UKF) to address the problem of instability in numerical computation. In addition, by setting an adaptive factor to optimize the MEE-UKF, an appropriate value of the KB can be obtained adaptively. The high accuracy and robustness of the AMEE-UKF were demonstrated by the simulation experiments.
展开▼
