Design of Fusion Filtering Method for Rectangular Descriptor Systems with Randomly Occurring Measurement Delays

In this paper, the rectangular descriptor systems with randomly occurring measurement delays are discussed and the related fusion filtering problem is considered, where the randomly occurring measurement delays are modeled by Bernoulli random variables. Firstly, matrix transformation, singular value decomposition method and augmentation technique are applied in turn to transform the original rectangular descriptor systems into a set of non-descriptor subsystems. Secondly, the local filter gains that minimize the filtering error covariances are given and the fusion filtering scheme for the original descriptor systems is given by applying the model transformation and covariance intersection (CI) fusion technique. Finally, a numerical simulation is used to show the effectiveness of the proposed fusion filtering criterion.