A new actuator activation policy for performance enhancement of controlled diffusion processes. (Q1428129)

The present work aims at the development of a framework within which a moving actuator activation policy and feedback controller synthesis are integrated for diffusion processes modelled by parabolic partial differential equations. It is assumed that the process of interest has either: (i) multiple actuators and the desirable arrangement is to activate only one while keeping the remaining ones dormant, or (ii) a single actuator capable of moving at a priori selected positions within the spatial domain. Practical advantages associated with arrangement (i) are energy cost savings and simplification of the local controller synthesis and the overall switching scheme, and with (ii) is enhanced spatiotemporal disturbance compensation. Feedback controller synthesis methods based on linear matrix inequality techniques are employed for a finite-dimensional Galerkin approximation of the original distributed parameter system. Along with standard controllability criteria, additional conditions are imposed that ensure robustness with respect to a certain class of disturbances. The value of a performance functional is then explicitly calculated by solving a location-parameterized family of Lyapunov matrix equations, and then optimized with respect to the set of admissible actuator locations. Finally, a case study of a moving bed adsorber is presented where the performance-enhancing capabilities of the proposed method is evaluated through simulation studies.