Identification Of Continuous Nonlinear Systems Research Articles

Wavelet neural networks for continuous time nonlinear system identification

Abstract In this paper, a wavelet-based neural network (WNN) is introduced for continuous nonlinear system identification. The adaptive weight learning laws are derived based on Lyapunov stability theory for static and dynamical system identification respectively. It is proved that the identification error will converge to zero and the weight will be bounded in the case of no modeling error. Simulation results demonstrate the effectiveness of the proposed identification methodology.

Nonparametric Identification of Continuous Nonlinear Systems Using Rajbman's Dispersional Method

The paper deals with a nonparametric, (statistical) identification model of continuous nonlinear dynamic system, especially with the identification of continuous Hammerstein systems with white noise input applying the dispersional method. It discusses how the Rajbman's cross dispersion function plays an analog role in the identification problem of continuous Hammerstein systems to that played by the cross correlation function in the case of the "active" identification of linear systems. The method introduced can ensure an optimal estimation of the nonlinear static and linear dynamic part of the Hammerstein system according to the mean square error. The paper gives results obtained by computer simulation as well as identification model applications.