Electronics
Machines à commutation de flux à grand nombre de phases : modèles comportementaux en mode dégradé et élaboration d’une stratégie de commande en vue de l’amélioration de la tolérance aux pannes
Publié le
In this thesis, we are interested in the study of a five-phase flux switching permanent magnet machine (five-phase FSPM machine) behavior in healthy and faulty mode. First, a comparison of electromagnetic performances between this machine and an equivalent three-phase machine is carried out. These performances are calculated by a Finite Element (FE 2D) model and validated by experiments. Results showed the five-phase machine contribution with a higher torque density, lower torque ripples, lower short-circuit current and ability to tolerate phases faults. The study of open-circuit tolerance is then developed for this five-phase FSPM. The behavior of the machine (the average torque, torque ripples, copper losses and the current in the neutral) in the case of open-circuit on a single and two adjacent and non-adjacent phases is presented. Then reconfiguration methods to improve the operation are proposed including a minimum reconfiguration allowing to end up with a feeding equivalent to that of a three-phase or a four-phase machine, an analytical calculation of optimal currents to cancel both the neutral current and torque ripples while ensuring the average torque, and finally a reconfiguration performed by a genetic optimization algorithm which is a non-deterministic algorithm multi-objective functions and multi-constraints. In this context, various combinations of different objectives and constraints are proposed and optimal currents are injected into the 2D FE model of the machine to see if performances have been improved. The analytical model of the torque used in the optimization algorithm is then revised to take into account the influence of the degraded mode. Different solutions of Pareto front are analyzed and electromagnetic performances are improved. This is verified by FE 2D calculations and followed by experimental validation. Faults impact on the radial magnetic forces is also analyzed. In the second part of this work, the study of the five-phase FSPM machine tolerance to short-circuit faults is performed. First steps of the faults isolation are proposed. Thereafter, short-circuit currents, taking into account the reluctance machine impact, are calculated analytically and their effects on machine performances are analyzed. Reconfigurations are also calculated by the genetic algorithm optimization and new references currents improved the degraded mode operation. All results are validated by the FE 2D calculation and experimentally. In conclusion, comparisons between fault-tolerance to phases openings and short-circuits of the five-phase FSPM machine are performed. Results led to conclude regarding the operation of this machine in healthy and degraded modes with and without correction. Analytical, numerical and experimental results showed good efficiency of the proposed control to improve fault-tolerance to phases openings and short-circuits.