Controller design from data under UBB noise

A Youla–Kucera formulation of the controller design from data problem

“…A different approach to solve the DDC problem follows a deterministic formulation using Set-membership techniques. Recent results on this approach can be found in [4], [5] and [6].…”

Limited-complexity controller tuning: A set membership data-driven approach

“…Set Membership (SM) estimation techniques, where noises are assumed as unknown-but-bounded signals, have been satisfactorily applied in system identification [11,12], filter design from data [13,14,15] and controller design from data [16,17,18,19]. The problem of data-driven controller tuning for linear systems has been investigated in [16,17]. In [17], the SM Errors-in-Variables (SMEiV) identification method is applied to solve the controller tuning problem.…”

Novel data-driven control approaches : application in essential oil extraction processes

“…SM estimation techniques have been satisfactorily applied in system identification Milanese and Novara (2004); Milanese and Taragna (2005), filter design from data (Milanese et al, 2010;Ruiz et al, 2010;Novara et al, 2013b). The problem of data-driven controller tuning for linear systems has been investigated in Cerone et al (2017) and Valderrama and Ruiz (2014) the latter as a precedent of this dissertation. In Cerone et al (2017), the SM Errors-in-Variables (SMEiV) identification method is applied to solve the controller tuning problem.…”

“…Convex relaxations are employed to solve the resulting polynomial optimization problems, leading to computationally demanding solutions and therefore limiting the amount of experimental data that can be considered, even for a reduced set of controller parameters. In Valderrama and Ruiz (2014), set over-bounding techniques are used to derive efficient linear programming problems from the original non-convex problem, allowing to manage larger data sets. In the non-linear framework also exist relevant studies, for instance in (Novara et al, 2013a;Tanaskovic et al, 2015) it was formulated a novel method to data-driven controllers tuning, main characteristics of such work are: (i) The authors developed theoretical framework for the stability analysis of non-linear feedback control systems, (ii) It is presented a technique for the direct design of a controller from data, (iii) Under some assumptions the closed-loop stability is guaranteed for a set of trajectories of interest, (iii) A drawback of the method is that all state variables are assumed to be measured and a feasible state trajectory is required as reference signal, generated for example by an expert human operator.…”

“…First, in Chapter 2, considering previous results by the author in Valderrama and Ruiz (2014) a setmembership approach to give solution to DDC design problem is proposed, our formulation is named Set-Membership tuning (SMT). In SMT, the Unknown But Bounded (UBB) noise structure affecting the Set-Membership approach (SISO plants).…”