Study on the decoupling identification method of linear dynamic and nonlinear friction for servo drive system

Aiming at the linear and nonlinear friction characteristics that servo drive system present, this paper brings up a novel identification approach, which decouples the servo system into a combination of a linear component and a nonlinear component in identification process. The linear and nonlinear components are firstly modeled respectively. Then the system is converted to a structure, which is a linear model plus a nonlinear feedback model. In order to avoid the effect of nonlinear friction and obtain the accurate model of linear component, two-set unidirectional non-zero-crossing velocity square waves are used to excite the system response characteristics. The respective differences of the input and output signals of these two-set square waves are utilized as the identification signals to estimate the parameters of the linear model. Based on the obtained linear model of the system, the steady-state velocity outputs are used to determine the magnitudes of Coulomb friction. Simulation and experiment results have both demonstrated that this approach can improve the identification accuracy and enhance the validity and reliability of the servo controller.