Abstract: In this study, a new dynamic state space model was established by using the polynomial predictive idea and state dimension extension. We call the new model extended model, which was established without the exact knowledge of the original state dynamics, i.e., we way use the proposed extended model to describe the state dynamics no matter we know original state propagation well or not. A correspondent Extended State Particle Filter (ESPF) was then presented based on the proposed extended model. In the ESPF, the sum of the extended particle weights was applied to test whether the filter is convergent or not. Simulation results demonstrate that the proposed method still works well while the existed Particle Filter (PF) diverges in the situations that the state dynamics are not known well.