On the Use of Prescribed-Time Super-Twisting for Fusing Multiple Redundant Noisy Measurements

Abstract

© 2024 IEEE.This paper deals with the robust prescribed-time nonlinear state estimation for fusing multiple redundant noisy sensor measurements. Firstly, we present the prescribed-time super-twisting algorithm (PT-STA), which is a recent modified version of the classical super-twisting algorithm that ensures robust convergence within a prescribed time. Subsequently, the PT-STA is used to design a prescribed-time nonlinear robust state observer for second-order systems subject to disturbances and uncertainties. The observer is then combined with an average-based sensor fusion strategy that further improve the overall estimates in terms of noise rejection and sensor fault tolerance. We introduce sensor fusion algorithm for considering the availability of multiple redundant measurements. The effectiveness of the proposed scheme is illustrated via numerical simulations of a perturbed damped pendulum, under different numbers of measurements and also considering sensor fault. We compare the results with those obtained using the conventional super-twisting observer. The results indicate robust convergence of estimation errors to the origin within the prescribed time, a reduction in measurement noise as the number of measurements is increased, and sensor fault tolerance.

Keywords