Switched gain differentiator with fixed-time convergence

Acceleration of estimation for a class of nonlinear systems in the output canonical form is considered in this work. The acceleration is achieved by a supervisory algorithm design that switches among different values of observer gain. The presence of bounded matched disturbances, Lipschitz uncertainties and measurement noises is taken into account. The proposed switched-gain observer guarantees global uniform time of convergence of the estimation error to the origin in the noise-free case. In the presence of noise our commutation strategy pursuits the goals of overshoot reducing for the initial phase, acceleration of convergence and improvement of asymptotic precision of estimation. Efficacy of the proposed switching-gain observer is illustrated by numerical comparison with a sliding mode and linear high-gain observers.