Integrate-and-Differentiate Approach to Nonlinear System Identification

In this paper, we consider a problem of parametric identification piece-wise linear mechanical system described by ordinary differential equations. We reconstruct the phase space investigated from accelerometer data and perform parameter using iteratively reweighted least squares. Two key features our study are as follows. First, use differentiated governing equation containing acceleration velocity main independent variables instead conventional in position. Second, modify squares method including an auxiliary reclassification step into it. The application allows us to improve accuracy through elimination classification errors needed for estimation Simulation Duffing-like chaotic experimental aluminum beam with asymmetric joint show that proposed approach is more accurate than state-of-the-art solutions.