Feed-Forward Guided Generalized Predictive Control of PMSM Drive

The aim of this paper is a design of a linear state controller of the PMSM drive. Popular generalized predictive control (GPC) approach can yield a simple controller, however it has difficulty with meeting hard constraints that are imposed on the drive (specifically, the drive must secure powerful limitation of motor current and keeps demand for ultimate drive dynamics). We propose to resolve this difficulty by smart design of set-points in the GPC criteria. Specifically, we design a feed-forward controller based on model of the drive that is capable to respect the above mentioned hard constraints. The result of this feed-forward controller is then used as a set-point for the GPC. Hence, with properly chosen penalization matrices, the GPC control operates only in close proximity of the set-point and improves over the feed-forward controller. The feed-forward control is based on a mathematical model of PMSM and the whole control strategy is designed to be as simple as possible. The theoretical conclusions are verified by experiments made on developed drive prototype of rated power of 10.7kW. The results of experiments show excellent behavior of proposed GPC based controller with feed-forward link and the data are compared with conventional PI based vector control. Furthermore, this paper also discusses and presents solution for optimization of proposed drive controller.