Modeling and Parameter Estimation of Electrical Multiphase Machines in Traction Applications
Time: Fri 2024-10-04 10.00
Location: E3, Osquars backe 18, Stockholm
Video link: https://kth-se.zoom.us/webinar/register/WN_6uOecoRNTGyRWn4ELR-7Ag
Language: English
Doctoral student: Gustaf Falk Olson , Elkraftteknik
Opponent: Dr. Franck Scuiller, Naval Academy Research Institute, École Navale Brest, France
Supervisor: Associate professor Luca Peretti, Elkraftteknik; Stefan Östlund, Elkraftteknik
The public defense is broadcast online. A link to the registration follows: https://kth-se.zoom.us/webinar/register/WN_6uOecoRNTGyRWn4ELR-7Ag
QC 20240911
Abstract
This thesis examines the modeling and parameter estimation of multiphase electrical machine (MPEM) and fault-tolerant permanent-magnet synchronous machine (PMSM) drives for traction applications. These drive systems typically require a wide constant power-speed range, substantial overload capability, and high efficiencies along with operational reliability.
A class of MPEMs known as variable phase-pole machines (VPPMs) has been identified as possible magnet-free candidates to fulfill these requirements. However, research on their modeling and subsequent parameter estimation is scarce. State-of-the-art models such as the vector-space decomposition (VSD)assume a fixed number of phases and half-wave symmetry of the stator current distribution. Therefore, the first part of this thesis proposes an alternative modeling through the harmonic plane decomposition (HPD). Using the HPD, we showcase reference generations for rotor field-oriented control to accomplish different non-trivial phase-pole configurations and a strategy for switching between them. A constrained loss-minimization also demonstrates that VPPMs can improve the overload capability and reduce losses compared to an identical machine operated with a fixed phase-pole number.
The second part of the thesis identifies the HPD parameters of a VPPM using standard tests performed on a reduced set of harmonic planes using solely a three-phase sinusoidal supply. This part shows that an optimally weighted least-squares estimation of the machine parameters can accurately determine the full suite of rotor resistances and unsaturated magnetizing inductances while segregating rotor and stator leakages. A key to doing so is to include a regularization term.
The third part of the thesis investigates inter-plane cross saturation. This phenomenon occurs between harmonic planes because the excited harmonics of the magnetic field share the same flux paths through the machine’s steel. In response, we propose an advanced Γ-model with saturable rotor bridge and stator inductances taking, respectively, the bridge and air-gap magneticflux densities as inputs. The result is saturation models with a single input implicitly accounting for the phase displacement and different amplitudes ofthe space-harmonic magnetic fields.
The fourth part of the thesis investigates a fault case of PMSMs termed active short-circuit (ASC) in which the machine terminals are deliberately clamped to a DC-rail. The resulting transient may cause irreversible demagnetization of the rotor permanent magnets and impose a sudden and appreciable braking torque on the machine shaft. From the non-linear flux maps of the PMSM, we deduce the set of initial conditions that do not demagnetize the permanent magnets and obey a user-defined torque limit. The model is used to elaborate a direct flux-vector-control based strategy to control the state variables into the safe operating area of initial conditions before applying the ASC. The strategy optimally uses the assigned voltage resources to decrease the stator-flux magnitude to an acceptable level within a predictable time. It is displayed and experimentally demonstrated how the strategy consequently protects the machine from a detrimental transient.
This work shows that appropriate modeling of VPPMs enables efficient use of the available input power and may extend the torque-speed operating region. It is also a prerequisite for accurate control in dynamic operation. While experimental parameter estimation of the HPD model can be made simpler and more robust using regularized estimators, dealing with magnetic cross-saturation poses a challenge when demanding constant and accurate torque output during pole-changing and harmonic injection. Currently, PMSMs remain commonplace in traction applications. Steps to improve their fault tolerance to ASCs are taken in this thesis, relieving a common concern in automotive applications that need to comply with functional safety standards such as ISO 26262.