Online parameter estimation in electric motors

Abstract: The permanent magnet synchronous machine (PMSM) can be used for a wide range of applications. In this case, the motor that is examined is inspired of Husqvarna's handheld electrical tools where the motor is part of its electrical chainsaw. The idea is to investigate and simulate different algorithms for online parameter estimation for the PMSM. Consequently, one of the algorithms is selected as the best one for this application. The motor parameters that are examined by the online parameter estimators are mainly the stator resistance and the inductance. The stator resistance is linked to the temperature of the motor and the stator inductance is also of interest for online estimation. Although the inductance is not directly linked to any physical quantities, large deviations deviations of the inductance usually indicate faults in the motor. By estimating these parameters, faulty operating points of the motor and critical damage can be stopped. The algorithms researched are; the recursive least square (RLS) with forgetting factor, extended Kalman lter (EKF) and the model reference adaptive system (MRAS). They are explained and applied to a non-salient PMSM, meaning that the stator inductance is not dependent on the rotor position. The simulations are done in Matlab/simulink with ten different test cases, emulating different operating points of the motors to see the convergence capability of the algorithms. The thesis also treats the subject of convergence time, and very lightly, the computational complexity of the algorithms. Moreover, throughout the simulation process, it is more evident that the EKF is the most suitable algorithm for online estimation in this application. This is because it handles all the different test cases well and converges to reasonable values within the margin of error.