Hybrid systems are a group of dynamical system which their behavior described by the interaction of discrete and continuous dynamical system behaviors. One of the subsets of hybrid systems, is piecewise affine system. Piecewise affine system identification, consists of estimating the parameters of each subsystem and the coefficients of the state-input boundary hyperplanes. In order to clustering the state-input space and estimating the feature matrixes simultaneously, bounded error algorithm and adaline neural network are used. It should be said that in this method, there is no need to know the number of linear subsystems of the piecewise affine system. Moreover, it should be noted that the identification method is extended based on on-line data acquisition from system. In continuation, this method is used to identify a benchmark mathematical piecewise affine system. By comparing the results with the reference paper, it is proven that this method has a good performance in clustering the state-input space and estimating the feature matrixes. In the end, by using the proposed method, an active water tank which its equations are described by the form of a piecewise affine system is identified.

Keywords: Hybrid Systems, Piecewise Affine System, System Identification, Bounded Error Algorithm, Adaline Neural Network

Full-Text [PDF 439 kb]   (5521 Downloads)