Estimation of model error for nonlinear system identification

Algorithms are presented for estimation of deterministic model error in the assumed models of nonlinear discrete and continuous time systems. The explicit model error time histories are parameterised using least squares method. The parameterised models relative to the true model explain the deterministic deficiency in the chosen models, in the sense of minimum model error. The algorithms have appealing features of extended Kalman filter. The numerical simulation results are obtained by implementing the algorithms in PC MATLAB