Model-Groups Design Methods for Multiple-Model Estimation with Variable Structure

Model-groups design is one of the most important problems in the application of the multiple-model (MM) estimator with a variable structure. The design of model-groups used in MM approach is considered in this paper. At first, two design algorithms for symmetric model-set are addressed by moment matching solution, if the true mode is a random variable and obeys certain probability distribution. Then the model-groups based on above designs are given, which give the practical guidance for use of model-set in MM approach and lead to improve the performance of MM estimation.