Iron Hysteresis and Enhanced Kalman Filtering for Sensorless Position Detection of Synchronous Motors

What I cannot create, I do not understand. Abstract T he auto-commutated permanent magnet synchronous motors, or brushless DC motors (BLDC), are more and more used in many different industrial low power (< 1 kW) applications. BLDC motors have the phase coils on the stator and the permanent magnet(s) on the rotor. Besides many advantages provided by this configuration , a drawback is that the position of the rotor has to be known in order to correctly drive the motor. Normally, Hall effect sensors or angular encoders are used for this purpose. In order to increase the reliability, reduce the size, manufacturing cost and wiring, a big research effort has been done for avoiding the use of direct sensors for the position detection of the rotor, leading to sensorless (or self-sensing if the motor is designed for that purpose) position detection. Effective solutions have been proposed for the nominal and high speed operations, as well as for the standstill position detection. However, in the case of BLDC motors with non-salient rotor, classical sensorless techniques using high frequency signal injections, state estimators or back-EMF detection still fail during the low speed operations. The objective of this thesis is therefore to propose new principles for the sensorless driving of BLDC motors with non-salient rotor, especially during low speed operations. Thanks to the introduced research, a deeper knowledge of the magnetic phenomena influencing the motor phase inductances and resistances is provided. It is highlighted that the phase inductances and resistances of most BLDC motors are naturally dependent on the rotor position. All known phenomena, like mechanical saliency, magnetic saturation, eddy currents, magnetic anisotropy, are analyzed and no one explains the measured inductance and resistance variations. Their origin is found in the non-linearity of the BH characteristic of the stator iron and, more accurately, in the local BH hystere-sis loops. Based on this knowledge, an existing BLDC motor is modified for increasing the effects of the highlighted phenomenon. Although small, the inductance and resistance variations are successfully used for detecting the rotor position. In order to maximize the information retrieval, a simple differential measurement is proposed and an innovative ready-to-use position-dependent vii Abstract signal is provided. On this basis, a first innovative sensorless position detection principle, naturally integrated into the classical two-phases ON supply, is introduced for the low speed operations. The principle is implemented in a prototype and extensively and successfully tested. An …