Self-optimizing Model Predictive Direct Torque Control for electrical drives

Model Predictive Control (MPC) offers a variety of advantages against linear control approaches (e.g. PI-Controller). Due to new approaches real-time implementation of MPC is already possible for permanent magnet synchronous motors with interior magnets (IPMSM). While the choice of the structure of the objective function to be minimized is quite intuitive, the adjustment of its weighting factors is not. In this paper a Model Predictive Direct Torque Control (MPDTC) approach is presented that automatically determines the weighting factors of the objective function. This self-optimizing MPDTC also offers real-time capablility for online MPC even with process time constants in the millisecond range. The good control dynamics (short settling time, low overshoot and small current ripple) as well as the differences between this self-optimizing MPC compared to the normal MPC is shown by simulation results.