Nonlinear Hammerstein model and parameter identification for servo drive system

Aiming at resolving the problem that linear models are not able to express the influence of the nonlinear phenomena in servo drive system, such as stick-slip friction, dead-zone etc, on condition of low-velocity or changing of directions, the Hammerstein model which contains static nonlinearity and dynamic linearity is adopted to describe servo system instead of linear model. The nonlinear characteristic of servo system is approximated by static nonlinearity, where two-segment polynomial is used to describe asymmetrical characteristics of friction. The recursive least square method is performed to estimate the parameters of model with multi-frequency sinusoidal input signal and velocity output signal of the servo system. The comparison of the model output between the linear model and the Hammerstein model in identification experiments indicates that the Hammerstein model can effectively represent the nonlinear characteristics of servo system, the error of the Hammerstein model output can be reduced by 90% of that of the linear model to greatly improving, the identification accuracy and precisely predicting the dynamic response of system in nonlinear states.