Abstract
The switching frequency of a PWM inverter feeding an induction motor drive is minimized in real-time. The algorithm predicts the future trajectories of the stator current space vector in the complex plane as functions of the possible inverter switching states. This information is used to select the optimal switching state. The computing time for the optimization is a few tens of microseconds. In order to eliminate such undesirable delay, a future state of the drive system is predicted which enables to perform the optimization in advance. This double prediction method renders the real-time minimization applicable to fast-switching PWM inverters.