Finite control set model predictive control (FCS-MPC), especially the reference-voltage-vector-based model predictive control (MPC) (RVV-MPC), considers a limited set of input vectors. However, the control performance is not satisfactory enough… Click to show full abstract
Finite control set model predictive control (FCS-MPC), especially the reference-voltage-vector-based model predictive control (MPC) (RVV-MPC), considers a limited set of input vectors. However, the control performance is not satisfactory enough due to a severe lack of voltage input variants. In particular, additional restraints, such as limiting the currents within setting boundaries, could not be supported with RVV-MPC. In this article, to truly comply with the boundary settings, a reference variant FCS-MPC approach is introduced. In the proposed strategy, reference currents are considered under the maximum torque per current (MTPC) control. As boundary settings could not be supported in RVV-MPC, it is better to search for the optimal current reference that is feasible within the imposed boundaries. Therefore, rather than fixing the current reference, a well-considered neighborhood of the MTPC reference is introduced, and these variants are processed, resulting in a set of optimal candidate inputs. As multiple, but still finite, optimal solutions become possible, additional efforts can be made within the objective function to restrain the current and reduce the torque ripple. The size of the current reference space can be automatically adapted to generate current references that correspond to truly bounded current vectors. A trade-off is then made between control performance and meeting the boundary settings.
               
Click one of the above tabs to view related content.