Study on wavelet repetitive control

Abstract

Repetitive control has been applied to robots, disk drives, etc. However, the memory size in repetitive controllers linearly increases with both the repetition time and the sampling frequency. Hence, the authors aim to apply discrete wavelet transforms to lower the memory requirements of repetitive controllers. The repetitive error signal is decomposed using a wavelet transform. Only the highest-level approximation coefficients and a few detail coefficients in the wavelet transform have to be saved and hence the error signal is compressed. The compressed error signal is in turn reconstructed to compensate the control system. Simulation results are presented to demonstrate the effectiveness of the proposed method.