Electrical and magnetic faults diagnosis in permanent magnet synchronous motors

Abstract

Permanent magnet synchronous motors (PMSMs) are an alternative in critical applications where high-speed operation,
 compactness and high efficiency are required. In these applications it is highly desired to dispose of an on-line, reliable and
 cost-effective fault diagnosis method. Fault prediction and diagnosis allows increasing electric machines performance and
 raising their lifespan, thus reducing maintenance costs, while ensuring optimum reliability, safe operation and timely
 maintenance. Consequently this thesis is dedicated to the diagnosis of magnetic and electrical faults in PMSMs.
 As a first step, the behavior of a healthy machine is studied, and with this aim a new 2D finite element method (FEM) modelbased
 system for analyzing surface-mounted PSMSs with skewed rotor magnets is proposed. It is based on generating a
 geometric equivalent non-skewed permanent magnet distribution which accounts for the skewed distribution of the practical
 rotor, thus avoiding 3D geometries and greatly reducing the computational burden of the problem.
 To diagnose demagnetization faults, this thesis proposes an on-line methodology based on monitoring the zero-sequence
 voltage component (ZSVC). Attributes of the proposed method include simplicity, very low computational burden and high
 sensibility when compared with the well known stator currents analysis method. A simple expression of the ZSVC is
 deduced, which can be used as a fault indicator parameter. Furthermore, mechanical effects arising from demagnetization
 faults are studied. These effects are analyzed by means of FEM simulations and experimental tests based on direct
 measurements of the shaft trajectory through self-mixing interferometry. For that purpose two perpendicular laser diodes are
 used to measure displacements in both X and Y axes. Laser measurements proved that demagnetization faults may induce
 a quantifiable deviation of the rotor trajectory.
 In the case of electrical faults, this thesis studies the effects of resistive unbalance and stator winding inter-turn short-circuits
 in PMSMs and compares two methods for detecting and discriminating both faults. These methods are based on monitoring
 and analyzing the third harmonic component of the stator currents and the first harmonic of the ZSVC.
 Finally, the Vold-Kalman filtering order tracking algorithm is introduced and applied to extract selected harmonics related to
 magnetic and electrical faults when the machine operates under variable speed and different load levels. Furthermore,
 different fault indicators are proposed and their behavior is validated by means of experimental data. Both simulation and
 experimental results show the potential of the proposed methods to provide helpful and reliable data to carry out a
 simultaneous diagnosis of resistive unbalance and stator winding inter-turn faults.