Extended Information Filter Based on Maximum Correntropy Criterion with Student’s t Kernel Function

This paper addresses the issue of information filtering for nonlinear systems with Student’s t noise interference. In the information filter framework, a new non-Gaussian extended information filter is developed under the MC criterion with Student’s t kernel function (TMCEIF). Firstly, the maximum correntropy (MC) filtering performance criterion with Student’s t kernel function (StKF) is defined. By comparing the Gaussian kernel function and adjusting the kernel bandwidth of the StKF, the StKF decays to zero more slowly for the large error, when the freedom degrees are small. Secondly, the MC criterion with the StKF is applied to design the new extended information filter (EIF). The robustness of the algorithm is verified by the final numerical simulation.