Compensation of nonlinear distortion effects for signal injection based sensorless control

D Salt, D Drury, DMJ Holliday

Research output: Chapter in Book/Report/Conference proceedingConference Contribution (Conference Proceeding)

4 Citations (Scopus)

Abstract

An accuracy and robustness comparison is made between three high-frequency carrier-based sensorless control techniques. The high-frequency carrier signal may be injected on an arbitrary stationary three-phase axis or the estimated, synchronously rotating d or q-axis. Nonlinear distortion effects introduced by the pulse-width modulation switching process lead to distortion of the high-frequency carrier signal which causes a degradation of position detection accuracy. It is shown that all three signal injection techniques suffer loss of accuracy but the distortion is least when the carrier is injected on the torque producing q-axis. This leads to a tradeoff between the undesirable production of high-frequency torque ripple and the radiation of audible noise, and robustness to nonlinear distortion effects. To overcome this tradeoff a new compensation model is proposed. An experimental implementation of the new compensator is applied to the three signal injection techniques used to estimate the rotor angle of a permanent magnet ac machine. Experimental results confirm that the new compensator minimises the distortion of the high-frequency injection signal irrespective of its axis of injection which leads to an improvement in position estimation accuracy and robustness.
Translated title of the contributionCompensation of nonlinear distortion effects for signal injection based sensorless control
Original languageEnglish
Title of host publicationPower Electronics, Machines and Drives, PEMD2010
Pages1 - 6
Number of pages6
DOIs
Publication statusPublished - 2010

Bibliographical note

Name and Venue of Event: Brighton, UK
Conference Organiser: IET

Fingerprint

Dive into the research topics of 'Compensation of nonlinear distortion effects for signal injection based sensorless control'. Together they form a unique fingerprint.

Cite this