In permanent magnet synchronous motor (PMSM) drive systems

sensorless control based on DC-link current sensing minimizes the number of sensors

reducing system cost and volume. However

low-speed sensorless control based on high-frequency (HF) signal injection disturbs DC-link current sampling

causing distortion in the reconstructed currents and degraded control performance. This paper proposes a HF signal injection strategy based on the measurement vector insertion method (MVIM). By dividing the switching period into modulation intervals and measurement intervals

the measurement interval remains unaffected by HF signal injection

ensuring accurate DC-link current sampling and phase current reconstruction. Furthermore

the current ripple and sampling delay error in MVIM are analyzed. To address the difficulties of using a phase-locked loop (PLL) for envelope extraction in HF injection methods

specifically

the challenges of parameter tuning and the trade-off between steady-state and transient performance

this paper proposes a three-state Kalman filter (TSKF) based on the Wiener process acceleration model. The TSKF provides optimal estimation of the states

enhancing both steady-state and transient performance in position observation

while requiring only a single parameter to be tuned. Experimental results validate the effectiveness and feasibility of the proposed MVIM and TSKF-based sensorless phase current reconstruction strategy.