Initial Position Detection and Polarity Identification for IPMSMs Using a Linear Time-Invariant Enhanced Phase-Locked Loop in HFSI

Xinran Shi, Jinlin Liu, Chao Gong, Jiasheng Yin

Research output: Contribution to journalArticlepeer-review

Abstract

This paper proposes an improved position sensorless control method for interior permanent magnet synchronous motors (IPMSM) using high-frequency square-wave injection (HFSI). Conventional HFSI methods obtain the initial position using a quadrature phase-locked loop (QPLL) and then identify the NS polarity through additional pulse voltage injection. This method requires extra signal injection, and the identification process is independent of the initial position estimation process, complicating the system design. To address these issues, a linear time-invariant enhanced phase-locked loop (LTI-EPLL) is introduced to replace the QPLL. The LTI-EPLL has a bidirectional convergence property, enabling it to estimate both the position and the amplitude of the input signal simultaneously. The estimated amplitude corresponds to the high-frequency response current amplitude in the HFSI, with its sign indicating the NS polarity. By using this method, the initial position and NS polarity can be determined simultaneously during high-frequency square-wave injection, significantly simplifying the system design process. This method also ensures accurate position estimation, contributing to high energy efficiency. The proposed LTI-EPLL-based HFSI method is verified through experiments on a 15 kW IPMSM.

Keywords

  • energy efficiency
  • high-frequency square-wave injection (HFSI)
  • initial position estimation
  • Interior permanent magnet synchronous motor (IPMSM)
  • linear time-invariant enhanced phase-locked loop (LTI-EPLL)
  • NS polarity identification
  • sensorless control

Fingerprint

Dive into the research topics of 'Initial Position Detection and Polarity Identification for IPMSMs Using a Linear Time-Invariant Enhanced Phase-Locked Loop in HFSI'. Together they form a unique fingerprint.

Cite this