Parameters identification for piecewise linear models with weakly varying noise.

This paper concerns the identification problem of piece–wise linear models from noisy data. The piece–wise linear models are of interest because they can approximate with arbitrary degree of accuracy any non–linear model, holding mathematical tractability and generality. The identification procedure is formulated within the Frisch scheme, well established for linear systems, which has been modified and improved to be applied for piece–wise linear systems. Further, using reasonable hypothesis on the data noise, the identification procedure is enhanced with respect to the linear case