Implementation and testing of Sigma-Point Kalman filters in Simulink for nonlinear estimation

Abstract

This thesis discusses the implementation of Sigma-Point Kalman Filters (SPKF) for state estimation of nonlinear wind turbine systems. First, a theoretical review of nonlinear Kalman filtering is given. Then the different ways of implementing the algorithms and testing them in SIMULINK are discussed and eventually the developed algorithms are explained and illustrative results from nonlinear simulations are presented. This work confirms that the linear Kalman Filter can be efficiently extended to nonlinear systems by means of Sigma-Point Kalman Filters such as the Unscented Kalman Filter and the Central Difference Kalman Filter. It is also shown that the performance of the square-root implementations available for SPKF is as accurate as that of the original ones, even if they are more computationally efficient algorithms.