Noise resistant design of wavelet based multiresolution control

Abstract

In this paper a new method for noise cancellation in a class of wavelet based multiresolution controllers is proposed. Noise cancelation is based on hard thresholding of detail signals in the decomposition stage of the controller. We took advantage of the fact that noise mainly affects high frequency contents of the error signal that is used by the controller, and used hard thresholding to filter their effect in the subsequent control output. The resulting controller outperforms conventional PID and multiresolution PID controllers in closed-loop feedback system design. We tested the performance of the new controller by implementing a n integral plus dead time process used integral of the time-weighted absolute error criteria (ITAE) to tune and compare different controller designs.