LAUSR

The ability to use actuator and load models online in combination with experimental data becomes increasingly important in future electrical drives. Among the applications of such online model in torque and flux controllers, model predictive control (MPC) has gained increasing interest, allowing to deal in real time and in a suboptimal way with drive variations captured by the model used. For such ability, sufficient processing power is required to handle the model in real time, which appears to be less of a problem in recent drives. This article contributes to the current prediction in MPC, applied to an interior permanent-magnet synchronous machine (IPMSM) drive. The study aims at improving the prediction, through an online estimation of the current time gradient coefficients, and this by considering consecutive space vector step responses. Through measuring the resulting current responses, an online correction of the current gradients is realized. A two-step method, which consists of an estimation step and a prediction step, will be described in order to improve the current prediction. Experimental results obtained from a 3.7-kW IPMSM drive will verify that the proposed geometrical approach effectively improves the conventional short window current predictions.