Contribution à la modélisation des défauts de bobinages des machines électriques et à leur détection par des mesures au rotor

Authors: Rouached, Bouali; Rouached, Bouali
Advisor: Cros, Jérôme; Cros, Jérôme; Clénet, Stéphane; Clénet, Stéphane
Abstract: This thesis deals with the detection of stator winding faults in electrical machines containing a damping structure. Short-circuits between turns of the same winding are often the premise for the appearance of other winding faults, the consequences of which are more severe. Early detection is therefore essential to preserve machines and their human or material environment. In this book, we propose an innovative solution to short-circuit discrimination based on the redundancy of signals measured at the rotor. To define this, high-order modelling combining the finite element approach and that of coupled circuits was used. The first is used to determine the parameters (magnetic couplings) exploited by the second which is none other than the behavioural model. As the downside of this combination of approach is the time taken to identify magnetic couplings and the knowledge that various coil configurations must be modelled, a new identification methodology was developed. After an initial identification (finite elements) at short notice, this offers the possibility of estimating the parameters of a desired winding configuration without reusing the finite element resolution. In addition to its use to implement the detection method presented, this identification methodology is a generator of useful parameters for detections based on "parametric estimation". The modelling of coupled circuits presents an excellent precision / calculation time ratio and a good efficiency in the representation of spatial-temporal phenomena. Integrating a winding fault requires changing the model's circuit equations. In order to achieve the objective of this thesis, the structure of the model has been modified to introduce the short circuits between turns at the stator. Apart from the application presented, the association of the identification method with the behavioural model allows the rapid production of multiple waveforms, particularly usable in "signal" oriented detection approaches. Finally, from the tool resulting from the previous work, the diagnostic process could be determined. The detection sensitivity is the main criterion for the definition of the instrumental technique and the signal processing of the process. The influence of the operating conditions on the detection sensitivity is evaluated and is considered for the choice of the solutions adopted. This process has been validated experimentally on a synchronous alternator with wound rotor of 5.4KVA.

This thesis deals with the detection of stator winding faults in electrical machines containing a damping structure. Short-circuits between turns of the same winding are often the premise for the appearance of other winding faults, the consequences of which are more severe. Early detection is therefore essential to preserve machines and their human or material environment. In this book, we propose an innovative solution to short-circuit discrimination based on the redundancy of signals measured at the rotor. To define this, high-order modelling combining the finite element approach and that of coupled circuits was used. The first is used to determine the parameters (magnetic couplings) exploited by the second which is none other than the behavioural model. As the downside of this combination of approach is the time taken to identify magnetic couplings and the knowledge that various coil configurations must be modelled, a new identification methodology was developed. After an initial identification (finite elements) at short notice, this offers the possibility of estimating the parameters of a desired winding configuration without reusing the finite element resolution. In addition to its use to implement the detection method presented, this identification methodology is a generator of useful parameters for detections based on "parametric estimation". The modelling of coupled circuits presents an excellent precision / calculation time ratio and a good efficiency in the representation of spatial-temporal phenomena. Integrating a winding fault requires changing the model's circuit equations. In order to achieve the objective of this thesis, the structure of the model has been modified to introduce the short circuits between turns at the stator. Apart from the application presented, the association of the identification method with the behavioural model allows the rapid production of multiple waveforms, particularly usable in "signal" oriented detection approaches. Finally, from the tool resulting from the previous work, the diagnostic process could be determined. The detection sensitivity is the main criterion for the definition of the instrumental technique and the signal processing of the process. The influence of the operating conditions on the detection sensitivity is evaluated and is considered for the choice of the solutions adopted. This process has been validated experimentally on a synchronous alternator with wound rotor of 5.4KVA.
Document Type: Thèse de doctorat
Thèse de doctorat
Issue Date: 2021
2021
Open Access Date: 7 June 2021
Permalink: http://hdl.handle.net/20.500.11794/69310
Grantor: Université Laval
Université Laval
Collection:Thèses et mémoires

Files in this item:
Description SizeFormat 
37192.pdf11.76 MBAdobe PDFThumbnail
View/Open
All documents in CorpusUL are protected by Copyright Act of Canada.