Implementation Aspects for State and Parameter Observers for Induction Machines at Low Sample-to-Fundamental Frequency Ratios

Abstract

The current developments in electric drive sys-tems are directed towards high rotational speeds and partial operation in the (thermal) overload range. This makes non-linear control theory increasingly common as the demands of high-performance drives increase, and high computing power becomes more accessible and available. However, high-speed drives are also confronted with a low sample-to-fundamental frequency ratio, which is particularly challenging for the observer design. In this case, the accurate timing of the measurements within the pulse-width modulation (PWM) pattern becomes relevant but is mostly neglected in the observer design. Therefore, this work investigates essential aspects of developing nonlinear state and parameter observers (extended Kalman filter) for induction machines, especially considering the timing of the PWM and current measurements. An observer is proposed, which systematically takes into account these effects. Simulation and measurement results prove that the proposed observer significantly improves estimation accuracy compared to the state of the art when electric drives with a low sample-to-fundamental frequency are considered.