Online parameter estimation and compensation of preisach hysteresis by SVD updating

Abstract

Preisach model can be used to describe the hysteresis in smart actuators. In this paper, online parameter identification by singular value decomposition (SVD) updating is presented for Preisach hysteresis. Firstly, SVD in least squares sense is used to obtain the pseudo inverse. However, it is time-consuming to compute singular values for large matrix, and PE condition is not always satisfied. Secondly, the SVD of large matrix is transformed to low order thin matrix, then, based on the initial result of offline SVD, online SVD updating is given by rotating basis which needs to be revised when new data is large. Finally, inverse Preisach feedforward is used to compensate the hysteresis. © 2011 IFAC.