The square root unscented Kalman filter formulation of risk-sensitive filter

Risk-sensitive filter is a robust and numerically efficient algorithm compared to risk neutral filter with model uncertainties. For nonlinear plant, the square root unscented Kalman risk-sensitive filter (SUKRSF) is proposed in this paper by using unscented transformation approximation. Square root unscented Kalman filter (SRUKF), a derivative-free nonlinear estimation tool is used to solve risk-sensitive problem because its several intrinsic properties suggest its use over extended risk-sensitive filter (ERSF) in highly nonlinear systems. The simulation results for certain nonlinear system also show that the new algorithm has better estimation performance than ERSF while driven under the identical machine model and parameters.