Bidirectional Helmholtz Machines

L'entraînement sans surveillance efficace et inférence dans les modèles génératifs profonds reste un problème difficile. Une approche assez simple, la machine de Helmholtz, consiste à entraîner du haut vers le bas un modèle génératif dirigé qui sera utilisé plus tard pour l'inférence approximative. Des résultats récents suggèrent que de meilleurs modèles génératifs peuvent être obtenus par de meilleures procédures d'inférence approximatives. Au lieu d'améliorer la procédure d'inférence, nous proposons ici un nouveau modèle, la machine de Helmholtz bidirectionnelle, qui garantit qu'on peut calculer efficacement les distributions de haut-vers-bas et de bas-vers-haut. Nous y parvenons en interprétant à les modèles haut-vers-bas et bas-vers-haut en tant que distributions d'inférence approximative, puis ensuite en définissant la distribution du modèle comme étant la moyenne géométrique de ces deux distributions. Nous dérivons une borne inférieure pour la vraisemblance de ce modèle, et nous démontrons que l'optimisation de cette borne se comporte en régulisateur. Ce régularisateur sera tel que la distance de Bhattacharyya sera minisée entre les distributions approximatives haut-vers-bas et bas-vers-haut. Cette approche produit des résultats de pointe en terme de modèles génératifs qui favorisent les réseaux significativement plus profonds. Elle permet aussi une inférence approximative amérliorée par plusieurs ordres de grandeur. De plus, nous introduisons un modèle génératif profond basé sur les modèles BiHM pour l'entraînement semi-supervisé.

Approaches to Provisioning Network Topology of Virtual Machines in Cloud Systems

The current infrastructure as a service (IaaS) cloud systems, allow users to load their own virtual machines. However, most of these systems do not provide users with an automatic mechanism to load a network topology of virtual machines. In order to specify and implement the network topology, we use software switches and routers as network elements. Before running a group of virtual machines, the user needs to set up the system once to specify a network topology of virtual machines. Then, given the user’s request for running a specific topology, our system loads the appropriate virtual machines (VMs) and also runs separated VMs as software switches and routers. Furthermore, we have developed a manager that handles physical hardware failure situations. This system has been designed in order to allow users to use the system without knowing all the internal technical details.

A real time electromagnetic analysis of electric machines for educational purposes and laboratory implementation

This thesis presents a system for visually analyzing the electromagnetic fields of the electrical machines in the energy conversion laboratory. The system basically utilizes the finite element method to achieve a real-time effect in the analysis of electrical machines during hands-on experimentation. The system developed is a tool to support the student's understanding of the electromagnetic field by calculating performance measures and operational concepts pertaining to the practical study of electrical machines. Energy conversion courses are fundamental in electrical engineering. The laboratory is conducted oriented to facilitate the practical application of the theory presented in class, enabling the student to use electromagnetic field solutions obtained numerically to calculate performance measures and operating characteristics. Laboratory experiments are utilized to help the students understand the electromagnetic concepts by the use of this visual and interactive analysis system. In this system, this understanding is accomplished while hands-on experimentation takes place in real-time.

Support vector machines to detect physiological patterns for EEG and EMG-based human-computer interaction:a review

Support Vector Machines (SVMs) are widely used classifiers for detecting physiological patterns in Human-Computer Interaction (HCI). Their success is due to their versatility, robustness and large availability of free dedicated toolboxes. Frequently in the literature, insufficient details about the SVM implementation and/or parameters selection are reported, making it impossible to reproduce study analysis and results. In order to perform an optimized classification and report a proper description of the results, it is necessary to have a comprehensive critical overview of the application of SVM. The aim of this paper is to provide a review of the usage of SVM in the determination of brain and muscle patterns for HCI, by focusing on electroencephalography (EEG) and electromyography (EMG) techniques. In particular, an overview of the basic principles of SVM theory is outlined, together with a description of several relevant literature implementations. Furthermore, details concerning reviewed papers are listed in tables, and statistics of SVM use in the literature are presented. Suitability of SVM for HCI is discussed and critical comparisons with other classifiers are reported.

Non-invasive winding fault detection for induction machines based on stray flux magnetic sensors

Non-intrusive monitoring of health state of induction machines within industrial process and harsh environments poses a technical challenge. In the field, winding failures are a major fault accounting for over 45% of total machine failures. In the literature, many condition monitoring techniques based on different failure mechanisms and fault indicators have been developed where the machine current signature analysis (MCSA) is a very popular and effective method at this stage. However, it is extremely difficult to distinguish different types of failures and hard to obtain local information if a non-intrusive method is adopted. Typically, some sensors need to be installed inside the machines for collecting key information, which leads to disruption to the machine operation and additional costs. This paper presents a new non-invasive monitoring method based on GMRs to measure stray flux leaked from the machines. It is focused on the influence of potential winding failures on the stray magnetic flux in induction machines. Finite element analysis and experimental tests on a 1.5-kW machine are presented to validate the proposed method. With time-frequency spectrogram analysis, it is proven to be effective to detect several winding faults by referencing stray flux information. The novelty lies in the implement of GMR sensing and analysis of machine faults.

Bidirectional Helmholtz Machines

L'entraînement sans surveillance efficace et inférence dans les modèles génératifs profonds reste un problème difficile. Une approche assez simple, la machine de Helmholtz, consiste à entraîner du haut vers le bas un modèle génératif dirigé qui sera utilisé plus tard pour l'inférence approximative. Des résultats récents suggèrent que de meilleurs modèles génératifs peuvent être obtenus par de meilleures procédures d'inférence approximatives. Au lieu d'améliorer la procédure d'inférence, nous proposons ici un nouveau modèle, la machine de Helmholtz bidirectionnelle, qui garantit qu'on peut calculer efficacement les distributions de haut-vers-bas et de bas-vers-haut. Nous y parvenons en interprétant à les modèles haut-vers-bas et bas-vers-haut en tant que distributions d'inférence approximative, puis ensuite en définissant la distribution du modèle comme étant la moyenne géométrique de ces deux distributions. Nous dérivons une borne inférieure pour la vraisemblance de ce modèle, et nous démontrons que l'optimisation de cette borne se comporte en régulisateur. Ce régularisateur sera tel que la distance de Bhattacharyya sera minisée entre les distributions approximatives haut-vers-bas et bas-vers-haut. Cette approche produit des résultats de pointe en terme de modèles génératifs qui favorisent les réseaux significativement plus profonds. Elle permet aussi une inférence approximative amérliorée par plusieurs ordres de grandeur. De plus, nous introduisons un modèle génératif profond basé sur les modèles BiHM pour l'entraînement semi-supervisé.

Design and application of experimental methods for steel sheet shearing

Shearing is the process where sheet metal is mechanically cut between two tools. Various shearing technologies are commonly used in the sheet metal industry, for example, in cut to length lines, slitting lines, end cropping etc. Shearing has speed and cost advantages over competing cutting methods like laser and plasma cutting, but involves large forces on the equipment and large strains in the sheet material. The constant development of sheet metals toward higher strength and formability leads to increased forces on the shearing equipment and tools. Shearing of new sheet materials imply new suitable shearing parameters. Investigations of the shearing parameters through live tests in the production are expensive and separate experiments are time consuming and requires specialized equipment. Studies involving a large number of parameters and coupled effects are therefore preferably performed by finite element based simulations. Accurate experimental data is still a prerequisite to validate such simulations. There is, however, a shortage of accurate experimental data to validate such simulations. In industrial shearing processes, measured forces are always larger than the actual forces acting on the sheet, due to friction losses. Shearing also generates a force that attempts to separate the two tools with changed shearing conditions through increased clearance between the tools as result. Tool clearance is also the most common shearing parameter to adjust, depending on material grade and sheet thickness, to moderate the required force and to control the final sheared edge geometry. In this work, an experimental procedure that provides a stable tool clearance together with accurate measurements of tool forces and tool displacements, was designed, built and evaluated. Important shearing parameters and demands on the experimental set-up were identified in a sensitivity analysis performed with finite element simulations under the assumption of plane strain. With respect to large tool clearance stability and accurate force measurements, a symmetric experiment with two simultaneous shears and internal balancing of forces attempting to separate the tools was constructed. Steel sheets of different strength levels were sheared using the above mentioned experimental set-up, with various tool clearances, sheet clamping and rake angles. Results showed that tool penetration before fracture decreased with increased material strength. When one side of the sheet was left unclamped and free to move, the required shearing force decreased but instead the force attempting to separate the two tools increased. Further, the maximum shearing force decreased and the rollover increased with increased tool clearance. Digital image correlation was applied to measure strains on the sheet surface. The obtained strain fields, together with a material model, were used to compute the stress state in the sheet. A comparison, up to crack initiation, of these experimental results with corresponding results from finite element simulations in three dimensions and at a plane strain approximation showed that effective strains on the surface are representative also for the bulk material. A simple model was successfully applied to calculate the tool forces in shearing with angled tools from forces measured with parallel tools. These results suggest that, with respect to tool forces, a plane strain approximation is valid also at angled tools, at least for small rake angles. In general terms, this study provide a stable symmetric experimental set-up with internal balancing of lateral forces, for accurate measurements of tool forces, tool displacements, and sheet deformations, to study the effects of important shearing parameters. The results give further insight to the strain and stress conditions at crack initiation during shearing, and can also be used to validate models of the shearing process.

Experimental study of strain fields during shearing of medium and high-strength steel sheet

There is a shortage of experimentally determined strains during sheet metal shearing. These kinds of data are a requisite to validate shearing models and to simulate the shearing process. In this work, strain fields were continuously measured during shearing of a medium and a high strength steel sheet, using digital image correlation. Preliminary studies based on finite element simulations, suggested that the effective surface strains are a good approximation of the bulk strains below the surface. The experiments were performed in a symmetric set-up with large stiffness and stable tool clearances, using various combinations of tool clearance and clamping configuration. Due to large deformations, strains were measured from images captured in a series of steps from shearing start to final fracture. Both the Cauchy and Hencky strain measures were considered, but the difference between these were found negligible with the number of increments used (about 20 to 50). Force-displacement curves were also determined for the various experimental conditions. The measured strain fields displayed a thin band of large strain between the tool edges. Shearing with two clamps resulted in a symmetric strain band whereas there was an extended area with large strains around the tool at the unclamped side when shearing with one clamp. Furthermore, one or two cracks were visible on most of the samples close to the tool edges well before final fracture. The fracture strain was larger for the medium strength material compared with the high-strength material and increased with increasing clearance.

Les machines du pouvoir. Technique et politique entre l'Iran sassanide et le califat abbasside

Dans l’Antiquité, la recherche sur la technique permet les premières réalisations de dispositifs ingénieux, tels que des appareils qui accomplissent une série d’actions par le biais de stimulus externes et de mécanismes cachés. Les organismes politiques et religieux saisissent rapidement la puissance communicative de ces machines, en devenant les promoteurs et patrons privilégiés de leur production. L’Empire sassanide (224-650) ne constitue pas une exception. En effet, les souverains perses consacrent, au moins à l’époque tardive, une grande attention à la conception et au déploiement de dispositifs savants. De même, un siècle plus tard, dans le milieu du califat islamique, les Abbassides (750-1258) semblent s’entourer de tels dispositifs. La continuité entre les deux empires dans plusieurs domaines, de la théorie politique à l’administration, est bien connue. Cependant, la question de la réutilisation du patrimoine technique et scientifique ancien, et notamment sassanide, par la cour abbasside, demeure encore largement inexplorée. L’étude d’un corpus de sources, aussi vaste qu’hétérogène, rassemblant des ouvrages historiographiques, géographiques, poétiques et d’adab, ainsi que des traités scientifiques et techniques en plusieurs langues, permet d’analyser différents aspects de la production et de l’usage politique des machines. Au sein de la cour sassanide, comme de la cour abbasside, la machine s’avère constituer un véhicule préférentiel de représentation et de diffusion de l’idéologie politique. À travers sa mise en scène publique, elle contribue de manière substantielle à la définition de l’espace du pouvoir, en participant à la création d’une image de la cour comme un microcosme au cœur duquel le Roi des rois, et plus tard le calife, occupaient le rôle cardinal de maître incontesté du monde. La continuité entre les empires sassanide et abbasside dans le domaine technique ne se limite donc pas à une récupération de savoirs, mais s’opère aussi sous la forme d’une véritable réactivation d’un patrimoine symbolique

Metal additive manufacturing of soft magnetic material for electric machines

This research work concerns the application of additive manufacturing (AM) technologies in new electric mobility sectors. The unmatched freedom that AM offers can potentially change the way electric motors are designed and manufactured. The thesis investigates the possibility of creating optimized electric machines that exploit AM technologies, with potential in various industrial sectors, including automotive and aerospace. In particular, we will evaluate how the design of electric motors can be improved by producing the rotor core using Laser Powder Bed Fusion (LPBF) and how the resulting design choices affect component performance. First, the metallurgical and soft magnetic properties of the pure iron and silicon iron alloy parts (Fe-3% wt.Si) produced by LPBF will be defined and discussed, considering the process parameters and the type of heat treatment. This research shows that using LPBF, both pure iron and iron silicon, the parts have mechanical and magnetic properties different from the laminated ones. Hence, FEM-based modeling will be employed to design the rotor core of an SYN RM machine to minimize torque ripple while maintaining structural integrity. Finally, we suggest that further research should extend the field of applicability to other electrical devices.

Hairpin windings for high reliability and high power density electrical machines

In the last years the increasing demand of higher torque and power densities has led to the adoption of hairpin windings (HWs) in electrical machines, mainly in those intended for automotive applications. However, this winding topology is quite sensitive to AC losses, hence one of their main challenges is represented by their reduction. This work deals with different design aspects related to the enhancements of some performance figures of rotating electrical machines for traction applications, above all power density and reliability, mainly through the adoption of HWs.

High power density electrical machines with hairpin windings

Nowadays, electrical machines are seeing an ever-increasing development and extensive research is currently being dedicated to the improvement of their efficiency and torque/power density. Compared to conventional random windings, hairpin winding inherently features lower DC resistance, higher fill factor, better thermal performance, improved reliability, and an automated manufacturing process. However, several challenges need to be addressed, including electromagnetic, thermal, and manufacturing aspects. Of these, the high ohmic losses at high-frequency operations due to skin and proximity effects are the most severe, resulting in low efficiency or high-temperature values. In this work, the hairpin winding challenges were highlighted at high-frequency operations and at showing the limits of applicability of these standard approaches. Afterward, a multi-objective design optimization is proposed aiming to enhance the exploitation of the hairpin technology in electrical machines. Efficiency and volume power density are considered as main design objectives. Subsequently, a changing paradigm is made for the design of electric motors equipped with hairpin windings, where it is proven that a temperature-oriented approach would be beneficial when designing this type of pre-formed winding. Furthermore, the effect of the rotor topology on AC losses is also considered. After providing design recommendations and FE electromagnetic and thermal evaluations, experimental tests are also performed for validation purposes on a motorette wound with pre-formed conductors. The results show that operating the machine at higher temperatures could be beneficial to efficiency, particularly in high-frequency operations where AC losses are higher at low operating temperatures. The last part of the thesis focuses on comparing the main electromagnetic performance metrics for a conventional hairpin winding, wound onto a benchmark stator with a semi-closed slot opening design, and a continuous hairpin winding, in which the slot opening is open. Lastly, the adoption of semi-magnetic slot wedges is investigated to improve the overall performance of the motor.

Novel prototypes of electro-activated molecular machines

The work presented in this thesis deals with the investigation of new prototypes of molecular machines, based on rotaxane and pseudorotaxane architectures, by means of voltammetric and spectroscopic techniques. The discussion is divided in two parts. Part I concerns the investigation of electro-switchable molecular muscles, based on mechanically interlocked molecules. This study is performed on systems of increasing complexity, starting from [2]rotaxanes and arriving to polymers. In Chapters 3 and 4, [2]- and [3]rotaxanes, characterized by the presence of three stations for the macrocycle(s), are investigated. In both systems, the macrocycle(s) movement can be controlled through a combination of stimuli, resulting in a processive and directional motion. In Chapter 5, daisy chain rotaxanes, dimers of the [2]rotaxanes discussed in Chapter 3, are investigated. These systems can be switched between an extended and a contracted conformation, and they represent the monomeric units for the realization of polymeric molecular muscles. In Chapter 6, the properties of electro-switchable polymeric molecular muscles, composed by the daisy chains investigated in Chapter 5, are discussed. The repeating units of these poly-daisy chains contract and extend upon electrical stimulation, and this motion is expected to be transmitted to the polymer itself, resulting in an amplification of the effect. Part II concerns the investigation of rotaxanes and pseduorotaxanes based on heteroditopic calix[6]arenes and cationic guests. In Chapters 8 and 9, novel calix[6]arene macrocycles, functionalized with thiourea or dansyl units, and their related pseudorotaxanes are investigated. In both cases, the calix[6]arene functionalization adds new features to the pseudorotaxane. In Chapters 10 and 11, the influence of orientational isomerism on the properties of [2]- and [3]rotaxanes is investigated. The [3]rotaxanes discussed in Chapter 10 display similar properties, while the [2]rotaxanes described in Chapter 11, characterized by a calix[6]arene and a stilbazolium unit, exhibit distinct photophysical and photochemical properties.

Diazene photoswitches in artificial molecular machines and light-effected materials

The research developed in this thesis focused on the spectroscopic and photochemical characterization of molecular diazene photoswitches, both as individual species and as functional components of mechanically interlocked molecules, molecular-based materials and artificial molecular machines and motors. Among the plethora of photochromes reported so far, azobenzene is the most versatile photoswitch due to its reproducible and well-established photochemical properties. Part I of this thesis work focuses on the characterization of light-responsive supramolecular systems based on azobenzene: a photochemically-driven rotary motor, a light-responsive supramolecular polymeric material and a supramolecular system capable of photoinduced entantiodiscrimination. Despite the wide success of azobenzene photoswitches, the tunability of their photochemical properties as a function of the diversified substitution pattern on its aryl ring presents intrinsic limitations. To overcome this issue, in the last decade heteroaryl azoswitches (i.e., azobenzene having heterocyclic rings in place of one or both phenyl groups) have attracted a great deal of attention. Hence, Part II of this thesis work treats the photochemical characterization of two different families of azoheteroarenes embedding imidazolium and thienyl functionalities in their structures. Their potential implementation in water-soluble artificial molecular machines and light-effected semiconductor materials is also assessed.