Tracking repetitive references and rejecting periodic disturbances are common tasks found in many control engineering applications. Repetitive Control (RC) is a learning control scheme that is designed to achieve both perfect accuracy tracking and good disturbance rejection. Recently, RC has been used in many applications, such as power supply control of proton synchrotron, disturbance rejection in disk drives, industrial robot control performing repetitive tasks, thickness control in cold rolling, active vibration compensation, attitude stabilization of satellites, total harmonic reduction in power inverter, disturbance rejection due to breathing in a flexible endoscopy system, and etc. RC is predictive control method which has a superior performance compared to the non-predictive control schemes. This is due to the capability of RC to learn the repetitive signal values and then generate them as control output. RC works well under the following conditions: (1) a constant periods of reference and disturbance signals and (2) the plant model is accurately known. However, if those conditions are not satisfied (e.g. the periodicity is varied and the plant model is time-varying or unknown), then the control performance will be significantly degraded (i.e. less accuracy and unstable system).

In this seminar, I will discuss RC-based control algorithms to overcome the limitations of conventional RCs. The proposed methods are named as: (a) Robust Repetitive Control (RRC), (b) Adaptive Repetitive Control (ARC), (c) Decentralized Repetitive Control (DRC), and (d) Sliding-Mode Repetitive Control (SMRC). I will also discuss applications that are used to verify the proposed methods and possible applications as my future works.