Proceedings of the 7 th International Conference on Applied Informatics Eger, Hungary, January 28–31, 2007. Vol. 2. pp. 359–366. Identification of dynamic systems by hinging hyperplane models Tamas Kenesei, Balazs Feil, Janos Abonyi University of Pannonia e-mail: abonyij@fmt.uni-pannon.hu Abstract This article deals with the identification of nonlinear dynamic systems by hinging hyperplane models, which are represented by tree structured piece- wise linear models. This type of non-linear black-box models is relatively new, and its iden- tification and application in the modeling of dynamic systems are not thor- oughly examined and discussed so far. They can be an alternative to artificial neural nets but there is a clear need for an effective identification method, because the original identication algorithm given by Breimann suffers from convergency and range problems. This paper presents a new identification technique based on a fuzzy clus- tering technique called Fuzzy c-Regression Clustering. This clustering tech- nique applies linear models as prototypes and the model parameters and fuzzy membership degrees are identified simultaneously. To use this cluster- ing procedure for the identification of hinging hyperplanes, there is a need to handle restrictions about the relative location of the hyperplanes: they should intersect each other in the operating regime covered by the data points. After the theoretical survey the paper gives detailed technical analysis of the proposed technique with the help of the identification of nonlinear process systems. Keywords: Hinging hyperplanes, fuzzy c-regression clustering, piecewise lin- ear models, dynamic models, regression tree 1. Introduction A lot of nonlinear regression techniques have been worked out so far (splines, artificial neural networks etc.). This article proposes a method for piecewise linear model identification applying hinging hyperplanes as linear submodels. Hinging hyperplane model is proposed by Breiman [3] and identification of this type of non-linear models is several times reported in the literature, because the original algorithm developed by Bremain suffers from convergency and range problems [8, 359