LAUSR

Hysteresis is a nonlinear effect that remarkably deteriorates the performance of smart actuators. To deal with this issue, numerous models have been proposed in the past several decades. However, to date, modeling hysteresis with sufficiently high accuracy is still challenging, since it is usually asymmetric, rate dependent, and amplitude dependent. Besides, how to efficiently convert the developed models into compensators for control purposes is another primary concern, which calls for novel solutions. In this article, we propose a unified approach to solve these problems. First, a feedback-assisted model inversion method, which functions regardless of the adopted model, is developed and tested under various conditions. Next, by fusing classical models and machine learning techniques into a disturbance estimation framework, it is found that both modeling and control performance can be significantly improved. Experimental validations are then conducted on a piezoactuated nanopositioner, demonstrating the effectiveness of the proposed method. As the developed model inversion technique requires minimum efforts for implementation, it thus can benefit a wide range of applications where hysteresis effect exists.