Model-set design method with representative points of probability distribution
基于概率分布代表点的模型集合设计方法

Multiple-model approach provides the state-of-the-art solutions to many problems involving estimation, filtering and control, etc. The design of a model-set is one of the important problems in the application of the multiple-model algorithm. A new method for designing model-set with representative points of probability distribution is proposed. Based on the given probability distribution of a system mode, the F-discrepancy or quasi F-discrepancy representative points and the mean square error representative points representing the probability distribution of system mode are selected by means of number-theoretic method. The model-set used in a multiple-model approach is then constructed. Model-set design in one-dimensional space and two-dimensional space is also given. Finally, simulation results show the performance of the model-set designed.