Probabilistic analytic center cutting plane method in robust H2 track following control

Abstract

The present paper addresses the design of discrete time robust H 2 track following dynamic output feedback controller for hard disk drives where uncertain parameters enter in a non-linear fashion into plant description. Uncertain parameters are considered as random variables with uniform distribution. The controller is designed to meet the performance specification with desired probabilistic levels (accuracy and confidence). The design is benefited from convex optimization in design parameter space and randomization in the uncertainty space. A localization method based on analytic center cutting plane algorithm is employed in order to find the probabilistic robust feasible solution. As a result of randomization, the computational complexity of the algorithm does not depend on the number of uncertain parameters and no conservatism is introduced while handling uncertain parameters. The effectiveness of the designed controller is verified through simulation as well as experiment. © 2013 Springer-Verlag Berlin Heidelberg.