Autonomous driving is increasingly popular among people and automotive industries in realizing their presence both in passenger and goods transportation. Safer autonomous navigation might be very challenging if there is a failure in sensing system. Among several sensing systems, image classification plays a major role in understanding the road signs and to regulate the vehicle control based on urban road rules. Hence, a robust classifier algorithm irrespective of camera position, view angles, environmental condition, different vehicle size & type (Car, Bus, Truck, etc.,) of an autonomous platform is of prime importance. In this study, Convolutional Neural Network (CNN) based classifier algorithm has been implemented to ensure improved robustness for recognizing traffic signs. As training data play a crucial role in supervised learning algorithms, there come an effective dataset requirement which can handle dynamic environmental conditions and other variations caused due to the vehicle motion (will be referred as challenges). Since the collected training data might not contain all the dynamic variations, the model weakness can be identified by exposing it to variations (Blur, Darkness, Shadow, etc.,) faced by the vehicles in real-time as a initial testing sequence. To overcome the weakness caused due to the training data itself, an effective augmentation technique enriching the training data in order to increase the model capacity for withstanding the variations prevalent in urban environment has been proposed. As a major contribution, a framework has been developed to identify model weakness and successively introduce a targeted augmentation methodology for classification improvement. Targeted augmentation is based on estimated weakness caused due to the challenges with difficulty levels, only those necessary for better classification were then augmented further. Predictive Augmentation (PA) and Predictive Multiple Augmentation (PMA) are the two proposed methods to adapt the model based on targeted challenges by delivering with high numerical value of confidence. We validated our framework on two different training datasets (German Traffic Sign Recognition Benchmark (GTSRB) and Heavy Vehicle data collected from bus) and with 5 generated test groups containing varying levels of challenge (simple to extreme). The results show impressive improvement by ≈ 5-20% in overall classification accuracy thereby keeping their high confidence.

Cette thèse traite de la conception d’un convertisseur DC / DC destiné aux véhicules électriques et hybrides (2,5 kW, 400V/14V, 250kHz). Dérivé de la topologie LLC à résonance, ce convertisseur bénéficie des nombreux avantages propres à cette structure particulière. C’est ainsi que le prototype réalisé présente un rendement très élevé, une densité de puissance très forte avec des perturbations EMI très réduites. La première partie de cette thèse est consacrée à l’analyse théorique du circuit LLC afin de dégager un modèle de conversion et une stratégie de contrôle adaptée à l’application visée. Afin de conserver un rendement important sur une large plage de charge, une structure basée sur la mise en parallèle de deux modules LLC est proposée. Une nouvelle stratégie de contrôle à deux boucles est également proposée pour équilibrer le courant entre les deux modules. La seconde partie de la thèse fait appel à la simulation et à l’expérimentation. Il s’agit de minimiser la masse et l’encombrement tout en maximisant le rendement. Un composant magnétique spécial est conçu puis dimensionné pour intégrer le transformateur et diverses inductances de résonance. Ce convertisseur met également en œuvre un système de redressement synchrone robuste avec une compensation de phase, un module de puissance avec une résistance thermique très faible et un système de refroidissement efficace par air. Le rendement maximal mesuré est 95%. Le rendement demeure supérieur à 94% sur une plage de puissance s’étalant de 500 W à 2 kW. La densité de puissance est 1W/cm3. La CEM du convertisseur est développée dans cette thèse. / In this dissertation, a 2.5kW 400V/14V, 250kHz DC/DC converter prototype is developed targeted for electric vehicle/hybrid vehicle applications. Benefiting from numerous advantages brought by LLC resonant topology, this converter is able to perform high efficiency, high power density and low EMI. A first part of this dissertation is the theoretical analysis of LLC: topology analysis, electrical parameter calculation and control strategy. To arrange high output current, this thesis proposes parallel connected LLC structure with developed novel double loop control to realize an equal current distribution. The second part concerns on the system amelioration and efficiency improvement of developed LLC. A special transformer is dimensioned to integrate all magnetic components, and various types of power losses are quantified based on different realization modes and winding geometries to improve its efficiency. This converter also implements a robust synchronous rectification system with phase compensation, a power semiconductor module, and an air-cooling system. The power conversion performance of this prototype is presented and the developed prototype has a peak efficiency of 95% and efficiency is higher than 94% from 500W to 2kW, with a power density of 1W/cm3. The CEM analysis of this converter is also developed in this thesis.

The purpose of this study was to compare 4 methods that assess the lower-body stretch-shortening cycle (SSC) utilization of athletes. Eighty-six National Collegiate Athletic Association Division I athletes from 6 different sports performed 2 squat jumps and 2 countermovement jumps on a force platform. Pre-stretch augmentation percentage (PSAP), eccentric utilization ratio (EUR), and reactive strength (RS) for jump height (JH) and peak power (PP) magnitudes, and reactive strength index–modified (RSImod) were calculated for each team. A series of one-way analyses of variance with a Holm-Bonferroni sequential adjustment were used to compare differences in PSAP, EUR, RS, and RSImod between teams. Statistical differences in RSImod (p < 0.001) existed between teams, whereas no statistical differences in PSAP-JH (p = 0.150), PSAP-PP (p = 0.200), EUR-JH (p = 0.150), EUR-PP (p = 0.200), RS-JH (p = 0.031), or RS-PP (p = 0.381) were present. The relationships between PSAP, EUR, and RS measures were all statistically significant and ranged from strong to nearly perfect (r = 0.569–1.000), while most of the relationships between PSAP, EUR, and RS measures and RSImod were trivial to small (r = 0.192–0.282). Pre-stretch augmentation percentage and EUR, RS, and RSImod values indicate that women's tennis, men's soccer, and men's soccer teams may use the SSC most effectively, respectively. Pre-stretch augmentation percentage, EUR, RS, and RSImod values may show vastly different results when comparing an individual's and a team's ability to use the SSC. Practitioners should consider using RSImod to monitor the SSC utilization of athletes due to its timing component.

Dans un contexte de mondialisation, les entreprises africaines doivent s'adapter en permanence aux diverses contingences économiques, juridiques ou sociales. Conscient de cette réalité, le législateur africain a mis en place dans le cadre de la reforme de l'OHADA, un arsenal juridique afin de leur permettre de s'acclimater en permanence aux tendances et pressions du marché : le droit de la restructuration qui désigne l’ensemble des techniques conçues par le législateur pour réorganiser l’entreprise. Ces techniques ont eu pour source d’inspiration le droit français des restructurations qui a connu des mutations. La confrontation des deux systèmes de droit permet de constater que si dans les principes fondamentaux les deux systèmes répondent aux mêmes objectifs à savoir assurer la pérennisation des entreprises par la promotion de la croissance de l’entreprise et sa survie lorsqu’elle est le seuil de difficultés diverses, il existe toutefois des différences d’ordre pratique et technique consécutives aux dernières réformes intervenues. / In a context of globalization, the african companies have to adapt themselves permanently to economic legal or social contingencies. Conscious of that reality, african legislator set up within the OHADA reform legal measures to allow theme acclimatize to the tendencies and markets pression : the restructuring law which design all the technicals proceedings conceived by the legislator to reorganize the company. Those technicals was inspirated by French law restructuring which has been reorganized. The confrontation of both systems allows to notice that if in the fundamental principles they are the same objectives consisting in substainability of companies for their growth or their survival there are however pratical and technical differences materialized by the recents french reforms.

Synaptic transmission alters the strength of the postsynaptic potential, through a process called short-term synaptic plasticity (STP). In this study, endplate potentials (EPPs) from the frog neuromuscular junction were used to resolve and quantify the presynaptic components involved in enhancement and depression of transmitter release during repetitive stimulation under normal quantal release conditions (2 mM Ca2+, 1mM Mg2+). During trains of stimulation given between 10 - 200 Hz, the amplitude of the EPPs first increased then decreased; a maximum increase of 77% was produced after 2-4 stimuli. EPP amplitudes began to increase at ~ 20 Hz, were maximal at ~ 55 Hz, and thereafter, decreased as the rate of stimulation increased. The integrated total release after 25 stimuli was little changed across frequencies between 10 - 100 Hz. EPPs ran down in two phases: a fast phase, attributed to the depletion of a readily releasable pool (RRP) of synaptic vesicles, followed by a slow phase, attributed to the depletion of vesicles from a depot pool (DP). Depletion of the readily releasable pool of synaptic vesicles (RRP) was determined by quantifying release under the fast and slow time rundowns and subtracting the number of vesicles associated with mobilization to the RRP from the total number of vesicles released during stimulation trains of 50 impulses. Impulses were delivered at 12 different rates ranging from 50 to 200 /s. Estimates of the number of vesicles released from the RRP increased with frequency of stimulation until maximal depletion levels of 5500 - 6000 vesicles were reached at stimulation rates between 90-130/s, assuming a control quantal content of 200 vesicles released per impulse. Depletion was less at lower frequencies when the number of stimuli delivered was identical. When the RRP maximally depleted, release was inversely related to stimulation rate, as would be expected if mobilization from the depot pool was the sole determinate of release during the slow phase. An equation constructed from four known components of enhancement and two components of depression - the depletion of vesicles from a readily releasable pool (RRP) and from the depot pool (DP) that refills the RRP, was used to fit and then simulate EPPs obtained during trains using different patterns of stimulation and varying amounts of extracellular Ca2+; the decay time constant parameters of enhancement, numerically derived from the observed data, were fixed at tau ~ 46, 220, 1600, and 20000 ms. The number of components of enhancement necessary to approximate the data decreased, from four in low (0.14 - 0.2mM) extracellular Ca2+, to one (tau ~ 46 ms) in 2.0 mM extracellular Ca2+, but four components of enhancement were necessary to fit the data when the amplitude of the EPP was not depressed below the control amplitude. This model was able to predict within ~ 3 % EPP amplitudes over a 10-fold range of frequency and Ca2+ concentration.

iv ABSTRACT ADAPTIVE CONTROL OF GUIDED MISSILES Tiryaki Kutluay, Kadriye Ph.D., Department of Aerospace Engineering Supervisor: Asst. Prof. Dr. Ilkay Yavrucuk February 2011, 147 Pages This thesis presents applications and an analysis of various adaptive control augmentation schemes to various baseline flight control systems of an air to ground guided missile. The missile model used in this research has aerodynamic control surfaces on its tail section. The missile is desired to make skid to turn maneuvers by following acceleration commands in the pitch and yaw axis, and by keeping zero roll attitude. First, a linear quadratic regulator baseline autopilot is designed for the control of the missile acceleration in pitch axis at a single point in the flight envelope. This baseline autopilot is then augmented with a Direct Model Reference Adaptive Control (DMRAC) scheme using Neural Networks for parameter estimation, and an L1 Adaptive Control scheme. Using the linearized longitudinal model of the missile at the design point, simulations are performed to analyze and demonstrate the performance of the two adaptive augmentation schemes. By injecting uncertainties to the plant model, the effects of adaptive augmentations on the linear baseline autopilot are examined. v Secondly, a high fidelity simulation software of the missile is used in order to analyze the performance of the adaptive augmentations in 6 DoF nonlinear flight simulations. For the control of the missile in three axis, baseline autopilots are designed using dynamic inversion at a single point in the flight envelope. A linearizing transformation is employed during the inversion process. These coarsely designed baseline autopilots are augmented with L1 adaptive control elements. The performance of the adaptive control augmentation system is tested in the presence of perturbations in the aerodynamic model and increase in input gain, and the simulation results are presented.

Ce travail est relatif à l’augmentation du courant de court-circuit du réseau électriqueportuaire alimentant le navire. Ceci passe par une modélisation des réseaux des charges, dutransformateur et enfin convertisseurs concernés. Munis de cette compréhension du système,diverses solutions à base d’électronique de puissance ou d’éléments passifs ont étéinvestiguées, simulées et évaluées à l’aune des performances exigées. Pour l’augmentation ducourant de court-circuit, un transformateur spécialement créé pour cette application à deuxenroulements primaires est proposé. L’esprit de la solution est de modifier la configuration dutransformateur au moment de la détection du court-circuit en modifiant son rapport detransformation. Un banc de test temps réel hybride a été architecturé afin de tester cettesolution. Les essais sur le banc de test ont permis d’obtenir des résultats satisfaisants etencourageants qui corroborent la théorie. Pour finir des études complémentaires, sousdifférents scenarii, ont été menées en simulation hors ligne. / This work is about the short-circuit current increase, inside the harbor’s electrical networkthat supplies the electrical grid of each vessel docked in port. We analyze the system’srequirements, the load demands, the transformer and finally the modeling of the convertersinvolved. Armed with this understanding of the system, various solutions like powerelectronics devices or passive equipments were investigated, simulated and evaluated in termsof performance requirements. To increase the short-circuit current, we provide a noveltransformer with two primary windings. The purpose of this solution is to change theconfiguration of the transformer when the short circuit is detected by changing thetransformation ratio. A real-time hybrid test bench has been build in order to test the solution.The real-time tests have yielded good and encouraging results, supporting the theory. Finally,further studies under various scenarios were conducted in offline simulations.

No description available.

La modélisation par équations structurelles est très répandue dans des domaines très variés et nous l'appliquons pour la première fois en métrologie dans le traitement de données de comparaisons interlaboratoires. Les modèles à équations structurelles à variables latentes sont des modèles multivariés utilisés pour modéliser des relations de causalité entre des variables observées (les données). Le modèle s'applique dans le cas où les données peuvent être regroupées dans des blocs disjoints où chaque bloc définit un concept modélisé par une variable latente. La structure de corrélation des variables observées est ainsi résumée dans la structure de corrélation des variables latentes. Nous proposons une approche bayésienne des modèles à équations structurelles centrée sur l'analyse de la matrice de corrélation des variables latentes. Nous appliquons une expansion paramétrique à la matrice de corrélation des variables latentes afin de surmonter l'indétermination de l'échelle des variables latentes et d'améliorer la convergence de l'algorithme de Gibbs utilisé. La puissance de l'approche structurelle nous permet de proposer une modélisation riche et flexible des biais de mesure qui vient enrichir le calcul de la valeur de consensus et de son incertitude associée dans un cadre entièrement bayésien. Sous certaines hypothèses l'approche permet de manière innovante de calculer les contributions des variables de biais au biais des laboratoires. Plus généralement nous proposons un cadre bayésien pour l'amélioration de la qualité des mesures. Nous illustrons et montrons l'intérêt d'une modélisation structurelle des biais de mesure sur des comparaisons interlaboratoires en environnement.

This thesis explores the methodology of state, and in particular, parameter estimation for time series datasets. Various approaches are investigated that are suitable for nonlinear models and non-Gaussian observations using state space models. The methodologies are applied to a dataset consisting of the historical lynx and hare populations, typically modeled by the Lotka- Volterra equations. With this model and the observed dataset, particle filtering and parameter estimation methods are implemented as a way to better predict the state of the system. Methods for parameter estimation considered include: maximum likelihood estimation, state augmented particle filtering, multiple iterative filtering and particle Markov chain Monte Carlo (PMCMC) methods. The specific advantages and disadvantages for each technique are discussed. However, in most cases, PMCMC is the preferred parameter estimation solution. It has the advantage over other approaches in that it can well approximate any posterior distribution from which inference can be made. / Master's thesis