Real-Time Control Reconfiguration for Active Disturbance Attenuation

In this work, the problem of active attenuation of disturbances with uncertain and possibly time-varying characteristics is addressed. The focus is on situations in which the uncertainty set is large, that is, the possible operating conditions of the system subject to disturbances can be different so that a single robust controller cannot ensure satisfactory performance levels. These situations underline the importance of control solutions able to reconfigure their action in real-time. The first part of this work is focused on the description of two different case studies, namely an active suspension system and an adaptive optics system, as well as on an overview of relevant contributions within the literature related to this subject. The solution that we propose is described in the second part; this method relies on the Adaptive Switching Control (ASC) paradigm, which has emerged as an alternative to classical Adaptive Control. In ASC, a finite family of candidate controllers is pre-synthesized off-line, and a supervisory logic has to select the potentially best one to be put in feedback with the plant. Particular attention is devoted to both performance and stability of the overall switching system. Finally, as an extension of the solution based on ASC, in the third part of this work an algorithm is proposed which combines both switching and tuning, aiming at preserving the beneficial features of the two different approaches, while possibly overcoming their drawbacks. The effectiveness of the proposed solutions is validated by means of simulation tests performed in the context of the two considered case studies.