Nonlinear Non-Gaussian Filtering Based on Divided Difference Filter and Approximate Conditional Mean Filter

In this paper, we present a new filtering algorithm for nonlinear dynamic state space models (DSSM) with non-Gaussian noise. The approximate conditional mean (ACM) filter has a better estimation performance for the DSSM with either the process noise or the measurement noise is non-Gaussian but not both, and the divided difference filter (DDF) has a better estimation performance for almost every nonlinear system under the Gaussian condition. On analyzing DDF and ACM filter, we developed a new ACM filter based on DDF, and it improved the performance of the tradition ACM filter. Experiments show that the proposed method works well in the filtering for DSSM with non-Gaussian noise.