Application of the Unscented Kalman Filter for the encoderless control of a permanentmagnet synchronous machine
The current rotor speed and the current rotor position are required for a field oriented control of a permanent magnet synchronous machine. Estimation of those parameters is difficult especially during low speeds. Position sensors are cost-intensive and a failure leads to an unstable operation of the motor. Encoderless control eliminates these drawback. The Unscented Kalman Filter is a special kind of observer that approximates the distribution of values with sigma points. With this observer the physical system is modeled for the estimation of rotor speed and the rotor position. Dynamic machine parameters from a FEM simulation are used to increase the modeling accuracy of the system.