Full Load Range Neural Network Efficiency Optimization of an Induction Motor with Vector Control using Discontinuous PWM

Model-based efficiency controllers of induction motors (IM) are known to considerably reduce power losses for loads under 0.50 p.u. by correctly setting the flux under rated value. However, for loads higher than 0.5 p.u., their performance is approximately the same as rated flux vector controllers. It has been recently shown that switching losses and harmonic losses reduction for voltage source inverter (VSI) fed IM can be done by using a discontinuous PWM strategy (DPWM). Hence, this paper proposes a full load range efficiency optimization controller by using an artificial neural network (ANN)-based flux optimizer and a DPWM strategy for both low and high loads. The controller is validated by using a simulation model and comparing its performance with a conventional rated flux controller with space vector PWM (SVPWM). Full load range optimization is achieved with a reduction of fundamental losses of 27% at light load and a 47% reduction of switching and time harmonics losses at high load