Estimation of Linear Systems with Abrupt Changes of the Noise Covariances Using Variational Bayes Algorithm

The variational Bayes method is applied to the state-space estimation problem with maneuvers or changes in the covariance of the observation noise. The resulting algorithm is an off-line batch method that can be used to provide a baseline performance estimation results for the recursive methods. In addition to batch methods we introduce a heuristic approach to make the algorithm on-line. Through simulations we show how the introduced method achieves the best accuracy out of all compared approximative estimation methods