Nonlinear Efficiency-Optimal Model Predictive Torque Control of
Induction Machines
Abstract
Induction machines (IMs) are widely used in many applications, e.g.,
electric vehicles or industrial automation, which motivates
efficiency-optimal operation for the sake of energy and cost savings.
However, a loss-minimal IM control leads to reduced flux operation at
partial load, i.e., achievable torque dynamics after load steps are
physically limited due to the IMâ\euro™s large rotor time constant.
Hence, designing the IM torque control for minimal settling time is of
particular importance to allow both an efficient and sufficiently
dynamic drive operation. Against this background, a model predictive
control (MPC) framework is proposed, which utilizes a precise model of
the IM covering magnetic saturation, iron losses, skin effect and
thermal influences. As this model is highly nonlinear, it is iteratively
linearized along the predicted control trajectory so that a
computationally efficient quadratic program can be defined for the MPC
problem. To enable sufficiently long prediction horizons, a hierarchical
control structure with the stator currents as the actuation variables of
the model predictive torque control is utilized. Thanks to this scalable
control approach, the proposed framework can be easily extended to
multi-machine drive systems covering higher-dimensional problem spaces.
Empirical tests validate the feasibility of the proposed approach both
for single as well as multi-machine drive applications.