Robust deadbeat control for synchronous machines rejecting noise and uncertainties by predictive filtering

A predictive current control scheme for permanent magnet synchronous motors (PMSMs) with deadbeat performance is presented. Major problems of such control schemes are the delay of the discrete control system, machine parameter uncertainties and measurement noise. With conventional methods, the delay can be easily compensated, but the effects of uncertainties can only be handled when they are comparably slow. This paper presents a new filtering method that attenuates the negative effects of parameter uncertainties in fast transients. The predictive controller inputs are actual measurements averaged with previous current predictions. As a result, the stability range is extended, and the oscillations caused by parameter uncertainties are damped. Noise sensitivity is also reduced. The system response is not slowed down. The results are confirmed analytically and experimentally.