Stator Current-Sensorless-Modulated Model Predictive Direct Power Control of a DFIM With Magnetizing Characteristic Identification
Shafiq Odhano, Sandro Rubino, Mi Tang, Pericle Zanchetta, Radu Bojoi
Abstract
This article presents a direct power control method based on modulated model predictive control for a doubly fed induction machine. The modulated predictive control algorithm constructs an optimal voltage vector from two inverter states that give a minimum absolute error in the active and reactive power. This article focuses on the effects of magnetic saturation and its impact on the accuracy of computed reactive power when the stator current sensors are not installed. To reduce the impact of magnetic saturation on reactive power computation, the machine's magnetizing characteristic is identified through a self-commissioning scheme introduced in this article. The identified magnetizing curve is utilized to construct a full-state stator flux observer, which is used to accurately estimate stator currents that appear in the reactive power equation. Experimental results are presented to demonstrate the accuracy of the reactive power computation in the absence of stator current sensors while conserving the rapid transient response offered by a modulated predictive control strategy for active and reactive power regulation.