Abstract:
This Thesis proposes a technique to control a nonlinear mechanical disk plant to follow a trajectory. A controller is designed based on feedback linearization and suppression of the system dynamics in discrete-time. The resulting closed loop system behavior is thus in algebraic form, i.e., the output in the future equals the current input with a constant time delay. In this way, the control of the output trajectory is easy. Compared to conventional feedback linearization, this technique does not require velocity and acceleration command. The experimentation on the disk plant demonstrates that the technique is able to control the plant to perfectly track a desired smooth trajectory. Compared to conventional feedback linearization, the proposed controller is more effective, especially when reference signal is unsmooth. In case that the system model contains unknown parameter and trajectory is periodic, a self-tuning ability is to be installed in the controller to estimate the unknown parameter. The purposed self-tuning algorithm is also successfully implemented to give a perfect tracking result even when the plant parameter is not precise at the beginning. In case that the reference trajectory contains a step or discontinuity, the trajectory should be modified such that it becomes smooth trajectory to avoid high control effort. The standard second order transfer function could be used to convert the step trajectory into a new smooth trajectory. The experiment demonstrates that this technique can effectively control the system to follow a step trajectory as well.