In this paper, we present a new algorithm for solving the LQG control problem with variance constraints which utilizes derivative information about the relevant ℋ2 costs to achieve quasi-Newton convergence. Using a lifting procedure, this algorithm is then generalized to work with linear periodically time-varying systems. This algorithm is then applied to the design of controllers for hard disk drives in order to assess the limits of performance of a particular setup. It is demonstrated that just by utilizing multirate sampling and actuation characteristics (i.e. without changing the hardware), the performance of this particular setup can improved by more than 39%.

This content is only available via PDF.
You do not currently have access to this content.