This paper deals with the robust power control of a grid-connected brushless doubly-fed induction generator (BDFIG) driven by the variable speed wind turbine. With the using of a super twisting algorithm which is a high-order sliding mode controller (HOSMC). This approach guarantees both the dynamic performance and the same robustness as traditional first order (SMC) algorithm and reduces the chattering phenomenon, which is the biggest disadvantage in the implementation of this technique. The developed algorithm relies on the decoupling control by implementing the strategy of oriented grid flux vector control. In order to enhance the desired performances, an attempt is made by controlling the generated stator active and reactive powers in a linear and decoupled manner to ensure the global asymptotical stability, HOSMC approach is implemented. Therefore, an optimal operation of the BDFIG in sub-synchronous operation is used in addition to the stator power flows where the stator power factor is kept in a unity. The suggested method is examined with the Matlab/Simulink software. The performances and the feasibility of the designed control are illustrated by simulation results.

Dans ce travail de thèse, nous avons proposé un système de supervision appliqué sur un robot manipulateur à deux degrés de liberté. La supervision est utilisée pour assurer la reconfiguration en temps réel du robot. Dans ce système nous avons utilisé une nouvelle méthode de détection de défaut (FD) de frottement visqueux du robot supervisé combinée avec un module de commande tolérante aux défauts (FTC).Le premier module, basé sur une méthode de traitement appliquée sur des résidus, va permettre la détection de défaut pour bien estimer les corrections nécessaires du deuxième module. Une évaluation de l’effet de défaut durant la supervision a été faite. Par ailleurs, le protocole TCP pour le transfert des données entre le robot superviseur et le robot supervisé a été utilisé. Les résultats de simulation montrent que la méthode proposée corrige l’effet de défaut en utilisant les données qui arrivent d’un robot superviseur à distance. Ensuite, nous avons proposé une implémentation matérielle sur cible FPGA de l’algorithme de supervision dont le but est de valider notre contribution et d’assurer un traitement en temps réel dans le cas où il y a des robots réels. Par ailleurs, une étude comparative entre les performances des deux implémentations a été effectuée

Le diagnostic est une composante principale du module de supervision. Il consiste à déterminer à chaque instant le mode de fonctionnement dans lequel le système se trouve. Il s‟appuie sur une connaissance à priori des modes de fonctionnement et sur une connaissance instantanée matériellement par une nouvelle observation de l‟état du système. Il existe plusieurs approches pour réaliser le diagnostic, le choix d‟une approche est lié au mode de représentation de la connaissance. Dans une première étape, nous allons essayer de faire la distinction entre les systèmes, les systèmes compliqués et les systèmes complexes, les différents types de systèmes. Nous introduisons ensuite la fonction de supervision, des systèmes industriels et des ses fonctions surveillance, détection et diagnostic, et les différentes approches de diagnostic

Located at the North-Eastern region of Algeria, the coastal aquifer of Taleza has a potential storage of groundwater. It contains hundreds of private wells installed by the population for several purposes including drinking water. Recently, the groundwater has become quite salinized. Furthermore, some people have given rise to the presence of a terrible odor, which smells as rotten eggs, in the groundwater pumped from their wells. Our first provided answer mentions that groundwater quality is mostly controlled by two factors: 1) the natural processes such as aquifer’s lithology, and soil/rock interactions with groundwater, and 2) anthropogenic activities such as agricultural production, industrial growth, urbanization with increasing exploitation of water resources, and private sanitation systems. However, the anthropogenic activities can be considered as the most serious sources of groundwater contamination. Unfortunately, most of Taleza’s people don’t have any idea about the anthropogenic groundwater contamination, and their raised question is « Is there still a chance to save the groundwater in Taleza aquifer? ». An urgent, efficient and durable groundwater protection planning is required in order to limit, and potentially improve the effects of the anthropogenic sources on groundwater quality within aquifer. In fact, determining the origin of saline components in aquifer of Taleza may provide more precision about the development of salinization process, and constitutes a useful tool to propose alternative solutions for ensuring the long-term protection of water resources. Accordingly, it is planned to perform a Multi-tracer study, combining isotopic and hydro-geochemical analysis, to identify the origin and fate of groundwater salinization. But firstly, the studying process will be started by reviewing the potential anthropogenic sources of groundwater contamination related to the Taleza territory. Our proposed communication describes a review of these potential anthropogenic sources, which potentially have led to the degradation of groundwater quality.

Recently, the identification of an individual race has become an important research topic in face recognition systems, especially in large-scale face images. In this paper, we propose a new large-scale race classification method which combines Local Binary Pattern (LBP) and Logistic Regression (LR) on Spark framework. LBP is used to extract features from facial images, while spark’s logistic regression is used as a classifier to improve the accuracy and speedup the classification system. The race recognition method is performed on Spark framework to process, in a parallel way, a large scale of data. The evaluation of our proposed method has been performed on two large face image datasets CAS-PEAL and Color FERET. Two major races were considered for this work, including Asian and Non-Asian races. As a result, we achieve the highest race classification accuracy (99.99%) compared to Linear SVM, Naive Bayesian (NB), Random Forest(RF), and Decision Tree (DT) Spark’s classifiers. Our method is compared against different state-of-the-art methods on race classification, the obtained results show that our approach is more efficient in terms of accuracy and processing time.