The precise information of initial rotor position is very important to high performance Interior Permanent Magnet Synchronous Machine (IPMSM) adopting incremental encoder, which can normally be obtained by the high-frequency (HF) rotating or pulsating voltage injection methods. However, the demodulation procedure may confront with the challenge of constrained stability and limited accuracy. For improving estimation accuracy, an improved impulse injection method with merits of simple and fast is proposed. It is carried out by generating a series of phase-axial injections to calculate the probable rotor position, then injecting two reciprocal voltage pulses to obtain the rotor polarity, and finally injecting iterative voltage vectors to obtain the real position rapidly. During the estimation process, the filters to extract the high frequency current signals are not needed. The method effectiveness is validated by the measured results of IPMSM test platform. In the same time, its application limitation is deduced by comparison with the measured results between IPMSM and surface-mounted PMSM (SPMSM). It is shown that the estimation algorithm is compatible with motor parameter differences and can reduce the influence of inductance saturation and nonlinear voltage error. Therefore, the proposed method not only improves the accuracy and robustness of the PMSM sensorless startup control, but also ensures the fast response.

To achieve better performance of sensorless control of PMSM, a precise and stable estimation of rotor position and speed is required. Several parameter uncertainties and variable measurement errors may lead to estimation error, such as resistance and inductance variations due to temperature and flux saturation, current and voltage errors due to measurement uncertainties, and signal delay caused by hardwares. This paper reveals some inherent principles for the performance of the back-EMF based sensorless algorithm embedded in a surface mounted PMSM system adapting vector control strategy, gives mathematical analysis and experimental results to support the principles, and quantify the effects of each. It may be a guidance for designers to minify the estimation error and make proper on-line parameter estimations.

The in-wheel electric vehicle with distributed drive units has better stability and flexibility than traditional centralized drives, but may encounter a higher failure rate due to additional actuators and sensors, especially that the faults of the wheel-side position sensor make motor torque out of control. To overcome this problem, a fault-tolerant control strategy with a multi-states switching method is proposed. The strategy judges the sensor failure by verifying redundant speed information, realizes sensorless control schemes by flux-observer based algorithm in high-speed range and I-F control algorithm in low-speed range with low acoustic noise, and applies adaptive transition process between different control schemes. To pursuit high stability, the signal-to-noise analysis for fault judgment due to sensorless estimation accuracy is discussed. Meanwhile, the principle of I-F resonant oscillations during the transition process is initially deduced in detail, and the conclusion of stability condition is obtained. Finally, the influence of system parameters on resonance performance is analyzed by simulation, and the effectiveness and reliability of the proposed strategy for the risk-controlling process are verified by experiments.

No abstract