Feedforward coefficient identification and nonlinear composite feedback control with applications to 3-DOF planar motor

Abstract

Due to modeling errors, accurate feedforward coefficient of the controller cannot be obtained with the standard method on the basis of the nominal model. Meanwhile, the system is uncertain in practice. Consequently, the MIMO (multi-input multi-output) system of the planar motor cannot be completely decoupled by feedback linearization, and the convergence of the tracking errors is no longer guaranteed. In order to improve the robustness and the tracking ability of the planar motor, a feedforward coefficient identification method and nonlinear composite feedback controller are proposed, thus guaranteeing stability by Lyapunov theory, wherein the feedforward coefficient can be obtained by the PD control experiment. The results of two different trajectory tracking experiments show that it is more accurate than the standard method. Moreover, this coefficient is suitable for different trajectories, so it avoids the drawback of ILC (iterative learning control) method, by which the feedforward term obtained cannot be reused if the length of the trajectory changes. The nonlinear composite feedback controller consists of u1 and u2 terms. u1 is designed to compensate for modeling errors, therefore the robustness is improved and the coupling effects among multi-DOF (degrees of freedom) are reduced. In balancing the trade-off between disturbance rejection and noise sensitivity, an amplitude-based variable-gain function is applied in u2. The trajectory tracking experimental results show that the overall controller is an attractive approach for the uncertain multi-DOF systems.