A Multiple Kernel Minimum Entropy Kalman Filter

In this paper, a multiple kernel minimum error entropy Kalman filter is proposed to estimate the filter state under abnormal measurement pollution. Firstly, an augmented Kalman filter model is constructed, and the minimum mean square error criterion under the traditional Kalman filter is replaced by the traditional minimum error entropy criterion. Then the quadratic rational kernel function and Gaussian kernel function are combined to construct the cost function, and the state results are obtained by fixed point iteration. Finally, the effectiveness of the filter is proved by a master-slave inertial navigation transfer alignment experiment.