Method of sensor noise attenuation in high-gain observers — Experimental verification on two laboratory systems

In this paper, a measurement noise attenuation technique is applied to a state observer in order to decouple the noise from the estimated state variables. It is important to limit such interferences, since the estimates are: 1) used online to formulate the state feedback controller, and 2) used in the compensation loop, where the estimated overall uncertainty of the system is partially canceled from the original plant. This measurement noise rejection method, which is based on extending the plant model with an additional, integral state variable, allows user to choose higher gains of the observer in order to provide faster convergence of its estimates, without the danger of making the system less energy-effective (or even destabilizing it) due to noise amplification. Experiments, conducted on two laboratory testbeds, show that the considered technique gives satisfactory results in practical application.