Ontwerp van regelaars met een vaste orde voor mechatronische systemen

The continuously increasing demands from industry drive research communities to push the limits in the design of accurate and high performance controllers for dynamical systems, such as autonomous vehicles, production machines, etc. Therefore, enhanced controller design procedures are indispensable in this evolution. Typically, first a mathematical model describing the behavior of a dynamical system is derived. Then, a controller using real-time measurements is designed for this model according to the desired performance specifications. For instance, a fast response should be guaranteed while limiting energy consumption. In a last step, the controller is validated in closed loop with the dynamical system. The complexity of a controller design problem depends on several factors, such as the dynamics of the system to be controlled, the desired controller structure, and the number/type of performance specifications. For linear time-invariant (LTI) dynamics, a wide range of successful controller design approaches have emerged over the last decades. However, several important control problems, such as the design of optimal fixed-order controllers (i.e., controllers with an a priori fixed structure), or optimal controllers satisfying multiple conflicting design objectives, remain unsolved. At the same time, many applications feature dynamics which are dominantly linear, but are affected by time-varying and/or uncertain parameters. In these applications, LTI controllers do not provide the desired accuracy and performance. Therefore, linear parameter-varying (LPV) dynamics have been widely considered as a useful extension, characterized by a linear input-output relation depending on real-time measurable parameters. Accounting for real-time parameter measurements in a controller design, higher closed-loop performance can be achieved compared to an LTI controller. Although modern LPV controller design techniques are very attractive when the system dynamics are affected by real-time measurable parameters, more involved approaches are required to cope with uncertain parameters. To meet the tightening performance and accuracy demands from industry, this thesis presents a versatile approach to design high performance fixed-order multi-objective controllers for the general class of linear parameter-dependent (LPD) systems, encompassing LTI, LPV and uncertain linear dynamics. For each of these subclasses, the effectiveness and practical viability of our approach is demonstrated by theoretical proofs of stability and performance, numerical comparisons with existing approaches, and experimental validations. In addition, a novel model order reduction technique is combined with our approach to design fixed-order controllers for continuous-time linear time-delay systems. Finally, a parametric programming approach is presented to design high performance feedback controllers for LTI systems, while simultaneously optimizing structural parameters affecting the system dynamics.status: publishe