Optimised implementation of adaptive GPC for low-order systems with time delays
In adaptive situations the comparatively large run-time computational burden and implementation complexity of the generalised predictive control (GPC) algorithm can prove troublesome. Presented is an optimised design to implement a GPC adaption mechanism for low-order systems. By adopting a compact state-space model and symbolically solving the underlying Riccati equations, an exact solution with a trivial code implementation is obtained. The proposed approach significantly reduces the computational and memory burden of the standard method, and gives improved performance over approximate schemes whilst exhibiting comparable overheads.
