A Modified k-plane Clustering Algorithm for Identification of Hybrid Systems

A new algorithm for the identification of discrete time hybrid systems in the piece-wise affine (PWA) form is introduced. This problem involves the estimation of both the parameters of the affine submodels and the partition of the PWA map from data. At the first stage we propose a modified version of the k-plane clustering algorithm proposed by Bradely and Mangasariang (2000) to provide initial data classification and parameter estimation. Then we apply the refinement algorithm proposed by Bemporad et al. (2003) repeatedly to the estimated clusters in order to improve both the data classification and the parameter estimation