Generalized Inversion of n-dimensional Flux Maps for Unified Nonlinear Machine Models and Predictive Control Algorithms

This paper describes a precise and fast method for inverting n-dimensional flux maps. These are necessary to fully model the nonlinearities of different machine types in a uniform and real-time capable way. With this tool, the known model predictive control methods (MPC) and simulation models for permanent magnet synchronous machines (PMSM) can be adapted to be used for a larger class of generalized machines, which includes electrically excited synchronous machines (EESM) and induction motors (IM) as well. Additional dependencies such as rotor position angle, temperature or speed can also easily be included in the map. This improves control quality and enables standardization of nonlinear controls across different machine types. In addition to that, the required memory to store such a flux map in lookup tables is significantly reduced by storing the flux map not in the dq-coordinate system, but in a coordinate system that is aligned to the magnetic couplings.