Nonlinear filtering for continuous-time systems using the linear fractional transformation model

In this paper, we propose Bayesian filtering technique for continuous-time dynamical models with sampled-data measurements using the linear fractional transformation (LFT) model which transforms the nonlinear state space model into an exact equivalent linear model with a simple nonlinear feedback loop. The linear model is amenable to Euler discretization. Simulation results demonstrate that the proposed filtering technique gives better approximation and tracking performance than the unscented Kalman filter (UKF) which diverges for highly nonlinear problems.