Position estimation method for self-sensing electric machines based on the direct measurement of the current slope

In self-sensing machines, the rotor displacement or angle is recovered from the wave form of the winding current, thus simplifying the structure and lowering the price of the machine. This presentation introduces a new technique to enable sensorless operation of a magnetic bearing or other self-sensing pulse-width modulation (PWM)-fed electric machines. The approach is based on the measurement of the rate of change of the winding current at the PWM frequency. The current slope along with the current and voltage measurements is used for estimating the winding inductance, which is a dynamic function of the rotor displacement and angle. The rotor state can then be deduced by matching the estimated inductance to the inductance model. The concept is tested in a setup in which the rotor of a switched reluctance machine is levitated against the gravity using feedback from a displacement sensor. The actual displacement is then compared to the estimate given by the self-sensing algorithm. The estimate is found to be accurate, but sensitive to modelling error.