Initial Position Estimation for IPMSMs Using Comb Filters and Effects on Various Injected Signal Frequencies

Abstract

This paper proposes fast initial position estimation of IPMSMs (interior permanent magnet synchronous motors) that is based on intentional pulse voltage injection and comb filters. Most initial position estimation methods utilize the amplitude or differential values of high-frequency currents, and the performance of sensorless control depends on the algorithm to reconstruct these values. Although Fourier series expansion, which requires some low-pass filters (LPFs), is utilized generally, the filters degrade the response of position estimation. This paper focuses on the algorithm for initial position estimation of IPMSMs at standstill, and proposes a new algorithm using comb filters that can rapidly calculate the amplitude values of high-frequency currents and improve the initial position estimation performance. In addition, this paper experimentally shows the performance of the initial position estimation, starting characteristics of sensorless control, and characteristics of sensorless control including standstill under the load.

Position sensorless starting method for IPMSMs by all‐pass filters

“…Here initial rotor position is assumed to be estimated using high-frequency signal injection. 13 The high-frequency waveform was 20 V, 1,250 Hz, and d-axis current (DC bias) passed to identify magnetic poles was 1.5 A.…”

“…In addition, Figure 12 shows results of initial rotor position estimation at standstill as proposed in reference. 13 In the diagrams, ω * rm , ω rm are speed commands, θ re , θˆr e are actual and estimated positions, and Δθ re = θˆr e -θ re is position estimation error. As can be seen from Figure 10, estimation performance remarkably declines where all initial vectors are set to zero, and a starting failure occurs.…”