Efficient Implementation of Controllers for Large Scale Linear Systems via Wavelet Packet Transforms

Abstract

In this paper we present a method of efficiently implementing controllers for linear systems with large numbers of sensors and actuators. It is well known that singular value decomposition can be used to diagonalize any real matrix. Here, we use orthogonal transforms from the wavelet packet to "approximate" SVD of the plant matrix. This yields alternatebases for the input and output vector which allow for feedback control using local information. This fact allows for the efficient computation of a feedback control law in the alternate bases. Since the wavelet packet transforms are also computationally efficient,this method provides a good alternative to direct implementation of a controller matrix for large systems.<p><I>This paper was presented at the 32nd CISS, March 18-21, 1998. </I>