Model predictive torque control of a Switched Reluctance Motor

Abstract

The strongly nonlinear magnetic characteristic of switched reluctance motors (SRMs) makes their torque control a challenging task. In contrast to standard current-based control schemes, we use model predictive control (MPC) and directly manipulate the switches of the dc-link power converter. At each sampling time a constrained finite-time optimal control problem based on a discrete-time nonlinear prediction model is solved yielding a receding horizon control strategy. The control objective is torque regulation while winding currents and converter switching frequency are minimized. Simulations demonstrate that a good closed-loop performance is achieved already for short prediction horizons indicating the high potential of MPC in the control of SRMs.