Tutorial: Methods to Identify & Control Highly Non-Linear Three-Phase Machines

Highly utilized three-phase machines show a highly nonlinear electromagnetic behavior, making it very challenging or even impossible to control them using standard control-algorithms. One very appropriate and well-proven method to cope with this nonlinearity is the measurement of multi-dimensional flux
linkage maps for each possible operating point of the given machine. During operation a look-up-table is used to adjust the gain of the used control-algorithm to the actual differential inductance in each given operating point. The flux maps are also used in non-linear model predictive control (MPC) schemes to
enhance dynamics. And besides machine-control the flux maps are implemented in high accuracy simulations to test new control algorithms. So with this method, the nonlinearities are stored in flux linkage maps and are fed-forward to the controller in each control cycle. To obtain the flux linkage maps, several methods are described in literature. One of the most-common methods is the steady-state method in which the device-under-test (DUT) is mounted in a hardware test-bench together with a load-machine. The load machine is speed-controlled and guarantees constant rotational speed, whereas the DUT is current controlled, enabling to drive it to every operation point in the dq-current plane. ... mehrThe downside to this method is the need of a real-power hardware test-bench, which is quite a cost factor and general effort. The other well-known method is the locked-rotor test in which the DUT’s rotos is locked and hence the rotational speed is zero. Here, no load machine is necessary but other restrictions apply, for example that no speed-dependent effects can be measured. In state-of-the-art implementations, the flux maps depend on the rotor-oriented direct and quadrature current components, considering the major nonlinearity-effects of magnetic saturation and cross-coupling. To be able to also consider nonlinearities that are due to the rotor and stator geometry, the dependency on the rotor angle must be taken into account as well. With these angle-dependent flux linkage maps, the angle-dependent error can be fed-forward e.g. in repetitive control schemes, enhancing control quality significantly. In this tutorial different methods to obtain multi-dimensional flux maps of permanent magnet synchronous machines (PMSM), synchronous reluctance machines (SynRM), electrically excited synchronous machines (EESM) and induction machines (IM) are presented. This includes steady state-tests, locked-rotor-tests, and a new approach that replaces flux maps with a physics-informed neural network. In addition to the flux-map-identification, also one well-proven control method that makes use of these flux maps and enables for high dynamics is
presented. Of course, also hands-on tips from our long-term lab-experience, dealing with several motor test-benches ranging from few hundred Watts (Pedelec/E-Bike motors) to several 100kW (automotive) for over a decade will be given in each of the described topics.