A modified super-twisting algorithm with specified settling time: numerical investigation

Resumo

The super-twisting algorithm (STA) is a finite-time stable algorithm that can be employed in control
and observation of dynamical systems to ensure a good transient performance of tracking and estimation errors.
However, only a conservative estimation of the convergence period can be obtained when using a classical STA.
To address this limitation, the present work proposes two novel specified-time stable algorithms, in which the
convergence instant can be directly specified as a system parameter. For the first one, we modify the right-hand
side of an STA by replacing its non-smooth continuous term with a time-varying specified-time stabilizing function.
For the second one, we alter the previously obtained system to recover the conventional STA performance after
the specified period. We extensively analyze the algorithms’ sensitivity to variations in each of their parameters
through numerical simulations. With proper tuning, the last proposed dynamic system is shown to provide robust
convergence to the origin at a specified instant of time.