Among the different multiphase ac drive solutions, one of the most widely reported in the literature is the six-phase machine. The machines can be realised into two different configurations, symmetrical and asymmetrical. For the symmetrical configuration, the stator winding consists of two sets of three-phase windings that are spatially shifted by 60 degrees where spatial displacement between any two consecutive phases is the same and equal to 60 degrees. For the asymmetrical configuration, the two sets of three-phase windings are spatially shifted by 30 degrees. As a result, the spatial shift between consecutive phases becomes non-equidistant. In this thesis, modulation techniques for both symmetrical and asymmetrical six-phase machines are investigated. The machines are configured in open-end winding configuration where both ends of the stator winding are connected to separate isolated inverters in a topology known as dual-inverter supply. Compared to conventional single-sided supply topology where one end of the winding is connected to an inverter while the other side is star-connected, some additional benefits are offered by the dual-inverter supply topology. First, fault tolerance of the drive is improved, since the supply is realised with two independent inverters. In case one of the inverters is faulted, the other can continue to provide power to the machine. Second, the same phase voltages can be achieved with half the dc-link voltages on the two inverter inputs compared to the single-sided supply, which can be useful in applications such as electric and hybrid electric vehicles and medium sized ships, where the dc voltage levels are limited. Further, due to the nature of the topology, additional diodes and capacitors like in the Neutral Point Clamped (NPC) and Flying Capacitor (FC) VSIs are not required. The latter results in a further advantage - capacitor voltage balancing techniques are not required. Two pulse width modulation (PWM) techniques for control of the dual-inverter supplied six-phase drives are proposed in this thesis. The first is a reference sharing algorithm where the inverters are modulated using reference voltage that is shared equally and unequally between the two modulators. For both symmetrical and asymmetrical six-phase drives, a better performance, in term of total harmonic distortion (THD) of phase voltage is obtained when the reference is shared unequally between the two modulators. The second technique is carrier-based modulation where the modulation of the two inverters is determined by the disposition of the carrier signals. Three variations of carrier signals disposition are investigated namely; the phase disposition (PD-PWM), alternate phase opposition disposition (APOD-PWM) and phase-shifted PWM (PS-PWM). For the symmetrical six-phase drive, the best phase voltage and current THDs are obtained using APOD-PWM while for asymmetrical six-phase drive, the APOD-PWM produces the worst current THD despite having the best voltage THD among the three methods. All the developed modulation techniques are analysed using simulations and experiments undertaken using a laboratory prototypes. The waveforms and spectra of phase voltage and load current obtained from the simulation and experimental works are presented in this thesis together with the THD of both the voltage and current over entire linear modulation range.

Knowledge of the end winding inductance of electrical machines is decisive for calculating their operating performance. In this article, two different approaches to analytically calculate the stator end winding inductance of large induction machines are discussed. The first method is based on the exact replication of the 3D conductor geometry using serially connected straight filaments, where the inductances are calculated by solving Neumann’s integral. In the second method, the end winding flux is resolved into components excited by the axial and circumferential end winding magnetomotive force, resulting in a far simpler geometrical model. In both cases, end face effects are taken into account by adopting the method of images. The analytical approaches are compared to the known analytical calculation method proposed by Alger [1]. In addition, the stator end winding inductance is computed by means of 3D finite-element analysis. Using experimental validation, it is shown that both the analytical and numerical results reasonably correlate with removed rotor inductance measurements taken for several induction machines with different rated powers and frame sizes, if the permeability of the laminated core is taken into consideration.

Grâce à leurs avancées techniques en termes de poids, performances et fiabilité, les machines synchrones sont de plus en plus utilisées dans le domaine de transport et en particuliers dans l’aéronautique. Les stratégies de maintenances de ces systèmes électriques sont alors indispensables afin d'éviter des surcouts liés à des temps d'arrêt non planifiés. Ce document propose une analyse des conséquences d'un défaut inter-spires intermittent naissant dans l'enroulement statorique d'une Machine Synchrone à Aimants Permanents (MSAP). Ce type de défaut correspond à l'état peu avancé d'un futur court-circuit permanent. Jusqu'à présent, les études menées se sont limitées à la détection de courts-circuits inter-spires permanents. L'objectif de cette analyse est de définir une méthode de détection de ce type de défaut qui soit facile à mettre en œuvre. A partir d'une étude analytique du courant statorique d'une MSAP contrôlée en courant en présence de court-circuit intermittent, nous avons étudié l'impact des différentes grandeurs influençant la perturbation du courant. Nous avons constaté que la forme de la perturbation créée par le défaut était toujours la même et qu'elle était la signature du défaut intermittent dans le courant. Par la suite cette étude analytique a été validée expérimentalement. Dans la partie suivante nous avons étudié la sensibilité des méthodes de détection des courts-circuits inter-spires permanents appliquées au cas du court-circuit intermittent. Ces méthodes se sont révélées inadaptées pour la détection du défaut étudié dans ce travail. Nous avons donc proposé une méthode dédiée qui est basée sur la détection de la signature du défaut et qui utilise une transformation en ondelette adaptée. Il s'agit d'une méthode de détection de forme qui permet non seulement de détecter le défaut intermittent mais aussi de le distinguer des autres types de défauts. La performance de la méthode a été validée par les résultats de simulation et de manipulation. Dans une dernière partie, une étude plus générale sur le suivi de vieillissement des enroulements est proposée. Elle est basée sur le suivi de l'évolution des courbes d'admittance hautes fréquences d'un bobinage au cours du temps en utilisant les fonctions de transfert de ce dernier. / Thanks to technical advances in terms of weight, performance and reliability, synchronous machines are increasingly used in the transport field and especially in aeronautics. The maintenance strategies of these electrical systems are essential to avoid extra costs associated with unscheduled downtime. This document offers a study on the intermittent inter-turn fault occurring in the stator winding of a Permanent Magnet Synchronous Machine (PMSM) and its consequences. This type of fault correspond to the emerging state for a future permanent short circuit condition. So far, studies have been limited to the detection of continuous inter-turn short circuits. The main purpose of this analysis is to define a detection method for this type of fault easy to implement. Based on the stator current analytical study of a PMSM current controlled in presence of intermittent short circuit, we had studied the impact of different variables influencing the current disturbance. We had found that the shape of the disturbance created by the fault was always the same and that it was the fault signature in the current signal. Later this analytical study was validated experimentally. In the next part we had studied the sensitivity of continuous short circuits detection methods applied in the case of intermittent short circuit. These methods have been proved unsuitable to detect the defect studied in this work. Therefore, we had proposed a dedicated method based on the fault signature identification using an adapted wavelet transform. It is a pattern detection method able to detect the intermittent fault and to distinguish it from other types of defects. The performance of the method was validated by simulation and experimental results. In the last part, a more general study concerning the winding health monitoring is proposed. It uses transfer functions and it is based on the monitoring over time of the winding high frequencies admittance curves evolution.

Multilevel inverters have advantages over two-level inverters such as reduced THD, ability to operate at low switching frequencies, reduced switching losses etc. Moreover, higher voltage levels can be handled with devices of lower voltage rating. The main disadvantage with the multilevel configurations compared to the two-level inverter configuration is the increase in the number of power devices required and the circuit complexity, which necessitates complex control schemes that add to the cost. Also, the reliability of the converters comes down as the number of devices increases. Reduction in complexity and modularity are desirable characteristics for the multilevel inverters. Open-end winding Induction Motor (IM) drive configurations are shown to have advantages over the motor drive schemes with isolated neutral. The DC-link requirement in case of open-end winding structures comes down to half the voltage rating of the conventional NPC inverters. The DC- link requirement in case of open-end winding structures comes down to half compared to that of the conventional NPC inverters. The number of switching states is higher in the case of open-end winding configuration compared to multiplicity of switching states of conventional NPC inverters, which gives a control flexibility that can be used for optimizing the hardware requirements. Taking advantage of the flexibility given by open-end winding configuration, this thesis proposes schemes which have reduced power circuit complexity. Non-sinusoidal voltage fed IM drives suffer from the problems related to the common mode voltage (CMV) generated by the inverters. This CMV causes bearing currents and shaft voltages which in turn cause increased conducted EMI, ground loop currents and premature bearing failure. A three-level scheme was proposed for an open-end winding Induction machine in the literature, which completely eliminate the CMV variation from the pole voltages as well as the phase voltages. This configuration uses 24 controlled switches and two isolated DC-sources. In this thesis, three-level inverter schemes with CMV elimination and reduced power device count for an open-end winding IM drive are proposed. The first scheme gets the reduction in switch count by sharing the top inverter of the three-level scheme and the second scheme achieves the same by sharing the bottom inverter. This way, the number of controlled switches comes down to 18 from 24. Another problem with multilevel inverters is the large number of isolated DC-sources required to achieve the multilevel inversion. Reducing the number of isolated supplies and using capacitors to split the voltage levels poses the problem of capacitor voltage balancing. A four-level inverter with both CMV elimination and capacitor voltage balancing for an open-end winding IM drive is proposed in this thesis. The motor is fed by two four-level inverters from both the sides. A closed loop capacitor voltage balancing scheme is implemented and the redundancies in the switching states are used for achieving the capacitor voltage balancing and thereby reducing the total number of DC-link to two. The control scheme is independent of the load power factor and maintains the balance in the entire modulation range. A five-level inverter scheme is proposed for an open-end winding IM drive in this thesis. It requires only two isolated DC-sources to achieve the five-level inversion. The motor is fed by one NPC three-level inverter from one side and a two-level inverter from the other. The inverters on either side share the DC-sources. Common mode voltage in the phases are made zero in an average sense using sine-triangle modulation in the proposed scheme so that the common mode currents through the phases are suppressed. The maximum fundamental voltage that can be obtained at the phase is limited to 0.5Vdc. DC-link requirement of the inverter scheme is half of that of conventional five-level inverter scheme because of the open-end winding structure. The two-level inverter, which should withstand half the DC-link voltage, is always in square wave operation and hence the switching losses are very less. All the schemes are simulated extensively in MATLAB/Simulink and experimentally verified on laboratory prototypes under V/f control. TI Motor control DSP and Xilinx CPLD/FPGA are used for generation of the PWM signals for the schemes. The inverters are switched at around 1.25 kHz to keep the switching losses low. Due to laboratory constraints, the experimental verification is done on low power prototypes. Nonetheless, the generality of the schemes allow them to be used for medium voltage high power applications.

Multilevel inverter technology has emerged recently as a very important alternative in the area of high-power medium-voltage energy control. Voltage operation above semiconductor device limits, lower common mode voltages, near sinusoidal outputs together with small dv/dt’s, are some of the characteristics that have made this power converters popular for industry and modern research. However, the existing solutions suffer from some inherent drawbacks like common mode voltage problem, DC-link capacitor voltage fluctuation etc. Cascaded multi-level inverter with open-end winding induction motor structure promises significant improvements for high power medium-voltage applications. This dissertation investigates such cascaded multi-level inverters for open-end winding induction motor drive with reduced switch count. Similar to the conventional two-level inverters, other multi-level inverters with PWM control generate alternating common mode voltage (CMV). The alternating common mode voltage coupled through the parasitic capacitors in the machine and results in excessive bearing current and shaft voltage. The unwanted shaft voltage may cross the limit of insulation breakdown voltage and cause motor failure. This alternating common mode voltage adds to the total leakage current through ground conductor and acts as a source of conducted EMI which can interfere with other electronic equipments around. As the number of level increase in the inverter, different voltage levels are made available by using DC-link capacitor banks, instead of using different isolated power supplies. The intermediate-circuit capacitor voltages which are not directly supplied by the power sources are inherently unstable and require a suitable control method for converter operation, preferably without influence on the load power factor. Apart from normal operation, the sudden fault conditions may occur in the system and it is necessary to implement the control strategy considering this condition also. A five-level inverter topology with cascaded power circuit structure is proposed in this dissertation with the strategy to eliminate the common mode voltage and also to maintain the balance in the DC-link capacitor voltages. The proposed scheme is based on a dual five-level inverter for open-end winding induction motor. The principle achievement of this work is the reduction of power circuit complexity in the five-level inverter compared to a previously proposed five-level inverter structure for open-end winding IM drive with common mode voltage elimination. The reduction in the number of power switching devices is achieved by sharing the two two-level inverters for both the inverter structures. The resultant inverter structure can produce a nine-level voltage vector structure with the presence of alternating common mode voltage. The inverter structure is formed by cascading conventional two-level inverters together with NPC three-level inverters. Thus it offers modular and simpler power bus structure. As the power circuit is realised by cascading conventional two-level and NPC three-level inverters the number of power diodes requirements also reduced compared to the conventional NPC five-level inverters. The present proposed structure is implemented for the open-end winding induction motor and the power circuit offers more number of switching state redundancies compared to any conventional five-level inverter. The inverter structure required half the DC-link voltage compared to the DC-link voltage required for the conventional five-level inverter structure for induction motor drive and this reduces the voltage stress on the individual power devices. The common mode voltage is eliminated by selecting only the switching states which do not generate any common mode voltage in pole voltages hence there will be no common mode voltage at the motor phase also. The technique of using the switching state selection for the common mode voltage elimination, cancels out the requirement of the filter for the same purpose. As the inverter output is achieved without the presence of common mode voltage, the dual inverter can be fed from the common DC-link sources, without generating any zero sequence current. Hence the proposed dual five-level inverter structure requires only four isolated DC supplies. The multi-level inverters supplied by single power supply, have inherent unbalance in the DC-link capacitor voltages. This unbalance in the DC-link capacitor voltages causes lower order harmonics at the inverter output, resulting in torque pulsation and increased voltage stress on the power switching devices. A five-level inverter with reduced power circuit complexity is proposed to achieve the dual task of eliminating common mode voltage and DC-link capacitor voltage balancing. The method includes the analysis of current through the DC-link capacitors, depending on the switching state selections. The conditions to maintain all the four DC-link capacitor voltages are analysed. In an ideal condition when there is no fault in the power circuit the balance in the capacitor voltages can be maintained by selecting switching states in consecutive intervals, which have opposite effect on the capacitor voltages. This is called the open loop control of DC-link capacitor voltage balancing, since the capacitor voltages are not sensed during the selection of the switching states. The switching states with zero common mode voltages are selected for the purpose of keeping the capacitor voltages in balanced condition during no fault condition. The use of any extra hardware is avoided. The proposed open loop control of DC-link capacitor voltage balancing is capable of keeping the DC-link capacitor voltages equal in the entire modulation region irrespective of the load powerfactor. The problem with the proposed open loop control strategy is that, it can not take any corrective action if there is any initial unbalance in the capacitor voltages or if any unbalance occurs in the capacitor voltages during operation of the circuit,. To get the corrective action in the capacitor voltages due occurrence of any fault in the circuit, the strategy is further improved and a closed loop control strategy for the DC-link capacitor voltages is established. All the possible fault conditions in the four capacitors are identified and the available switching states are effectively used for the corrective action in each fault condition. The strategy is implemented such a way that the voltage balancing can be achieved without affecting the output fundamental voltage. The proposed five-level inverter structure presented in this thesis is based on a previous work, where a five-level inverter structure is proposed for the open-end winding induction motor. In that previous work 48 switches are used for the realization of the power circuit. It is observed that all the available switching states in this previous work are not used for any of the performance requirement of CMV elimination or DC-link voltage balancing. So, in this proposed work, the power circuit is optimized by reducing some of the switches, keeping the performance of the inverter same as the power circuit proposed in the previous work. The five-level inverter proposed in this thesis used 36 switches and the number of switching states is also reduced. But, the available switching states are sufficient for the CMV elimination and DC-link capacitor voltage balancing. The advantage of the modular circuit structure of this proposed five-level inverter is further investigated and the inverter structure is modified to a seven-level inverter structure for the open end winding induction motor. The proposed power circuit of the seven-level inverter uses only 48 switches, which is less compared to any seven-level inverter structure for the open end winding induction motor with common mode voltage elimination. The power circuit is reduced by sharing four two-level inverters to both the individual seven-level inverters in both the sides of the of the open end winding induction motor. The cascaded structure eliminates the necessity of the power diodes as required by the conventional NPC multilevel inverters. The proposed seven-level inverter is capable of producing a thirteen-level voltage vector hexagonal structure with the presence of common mode voltage. The common mode voltage elimination is achieved by selecting only the switching states with zero common mode voltage from both the inverters and the combined inverter structure produce a seven-level voltage vector structure with zero common mode voltage. The switching frequency is also reduced for the seven-level inverter compared to the proposed five-level inverter. The advantage of this kind of power circuit structure is that the number of power diode requirement is same in both five-level and seven-level inverters. Since there is no common mode voltage in the output voltages, the dual seven-level inverter structure can be implemented with the common DC-link voltage sources for both the sides. Six isolated power supplies are sufficient for both the seven-level inverters. The available switching states in this proposed seven-level inverter are further analysed to implement the open loop and closed loop capacitor voltage balancing and this allow the power circuit to run with only three isolated DC supplies. All the proposed work presented in this thesis are initially simulated in SIMULINK toolbox and then implemented in a form of laboratory prototype. A 2.5KW open end winding induction motor is used for the implementation of these proposed works. But all these work general in nature and can be implemented for high power drive applications with proper device ratings.