Observer Design for Non-Uniform Sampled Systems Using Gain-Scheduling
In non-uniform sampled systems, the measurements are arriving at irregular time intervals. However, the control is updated at regular time intervals. An observer is required to obtain the estimate of the states during the control update times. We evaluate two observer designs: A Kalman filter and a gain-scheduling observer. The Kalman filter has the optimal performance. However, it is computationally expensive. In contrast, a recent gain-scheduling synthesis technique [1] can be used to design a time varying observer, whose time varying coefficients are a function of the measured sampling time variations. This observer is suboptimal, but it has significantly less computational complexity as compared to the Kalman filter, which makes it feasible to implement. Simulations are conducted for a self servo writing process in hard disk drives, in order to evaluate performance of H2 gain-scheduling observer design.