New Approach for Optimizing Inverter Losses in Finite Control Set Model Predictive Control of Induction Machine, Experimental Evaluation

In this paper a Finite Control Set Model Predictive Controller (FCS-MPC) is developed and implemented in real time on an FPGA platform. The common reference tracking squared l2-norm cost function is redefined in order to reduce the inverter losses directly instead of reducing the switching frequency. Different combinations of switching and conduction losses are considered in the proposed cost function aiming at more effective reduction of inverter losses. Experimental results show that, with the proposed cost function the switching operations happen more frequently when the corresponding phase current is small and avoided when it is high. This results in less inverter losses with the same average switching frequency compared to common cost function as well as lower DC-Link current ripple. In other words, with the same inverter losses, the proposed cost function has higher average switching frequency resulting in better current THD and torque TDD and consequently lower Induction Machine (IM) losses. The new cost function requires however, more FPGA resources.