State Estimation of Hybrid Data-driven Control System via Zonotopic Bounding Set Computation

This paper presents a data-driven method to obtain state estimation for a class of hybrid control systems, which involves interacting discrete-event and continuous-variable dynamics. But the system with no well-established model has made parameter optimization of the controller cannot be performed. The method relies on the virtual reference feedback tuning (VRFT) that ongoing direct use of the measurement I/O data and select desired controllers by employing output-error parameter estimation from virtual reference approximation. To improve the efficiency to compute a feasible state set that contains the parameters consistent with the measured data and the given noise and perturbation, a set membership estimation by zonotopes is used to implement the developed method. Besides, the effectiveness of the method is demonstrated experimentally on a leader-following tracking task of multiple unmanned aerial vehicles (multi-UAVs) formation flight.