Injectionless Acoustic-Noise-Based PMSM Sensorless Control Using RCPLL for Improved Estimation Performance

This article introduces an injectionless approach for acoustic-noise-based sensorless control of the permanent magnet synchronous machine. Replacing the high-frequency injection, the pulsewidth modulation (PWM) voltages are used as the basis for the method implementation. The mathematical model of P...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:IEEE transactions on power electronics 2024-01, Vol.39 (1), p.225-235
Hauptverfasser: Malekipour, Amirhossein, Corne, Adrien, Garbuio, Lauric, Granjon, Pierre, Gerbaud, Laurent
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:This article introduces an injectionless approach for acoustic-noise-based sensorless control of the permanent magnet synchronous machine. Replacing the high-frequency injection, the pulsewidth modulation (PWM) voltages are used as the basis for the method implementation. The mathematical model of PWM with considering the deadtime effect is first presented. Then, its impact upon the harmonics of current, magnetic fields, and generated radial forces is detailed. It is experimentally verified that in the presence of deadtime, unlike the contaminated current profile, there are merely specific-order speed-dependent components appeared in the acoustic noise spectrum, which is a clear bonus for signal processing precision and estimation performance. Moreover, a novel phase-locked loop (PLL) topology comprising a resonant-controller-based feedforward compensation block (RCPLL) is proposed. This compensation block is integrated within the PLL, by which the extra harmonics in the PLL error are directly suppressed, and therefore, the estimation performance is noticeably enhanced. The robustness of the proposed sensorless algorithm, alone, is verified experimentally under different conditions, and then, the estimation performances with and without the proposed RCPLL are compared.
ISSN:0885-8993
1941-0107
DOI:10.1109/TPEL.2023.3319824