DSpace University Ghardaia
DSpace is the repository platform that enable to share scientific product
Communities in DSpace
Select a community to browse its collections.
- Faculté des Sciences Economiques, Commerciales et des Sciences de Gestion
- Faculté de Droit et des Sciences Politiques
- Faculté des Lettres et des Langues
- Faculté des Sciences de la Nature et de la Vie et des Sciences de la Terre
- Faculté des sciences et technologies
Recent Submissions
Adaptive Traffic Signal Control Using Reinforcement Learning and Simulation
(university of ghardaia, 2026) MEHAYA Fatima; ABDELHADI Imane
Urban traffic congestion remains a critical challenge for modern cities, exposing the fundamental limitations of conventional fixed-time signal controllers. This thesis proposes and
evaluates a fairness-aware reinforcement learning (RL)framework for adaptive traffic signal
control, built around the Proximal Policy Optimisation (PPO) algorithm and implemented
within the SUMO microscopic traffic simulator.
A multi-objective reward function is designed to jointly optimise traffic efficiency and directional fairness, combining delay reduction, queue minimisation, throughput maximisation,
and Jain’s Fairness Index. Three distinct weight configurations are evaluated: efficiencyoriented, fairness-oriented, and balanced. Five controllers are systematically compared FixedTime, Vanilla PPO, PPO Efficiency, PPO Fairness, and PPO Balanced across two traffic
scenarios: a familiar Training Route (with static imbalanced demand) and an Unseen Route
(with temporally-varying symmetric demand) for zero-shot generalisation assessment.
Experimental results on the Training Route demonstrate that all PPO-based controllers
reduce average waiting time by 40–47% relative to the Fixed-Time baseline. PPO Efficiency
achieves the highest training throughput among RL controllers (655 vehicles) and the lowest
waiting time (15.63 s). PPO Balanced provides the most consistent multi-objective performance, achieving the lowest average queue length (168.2 vehicles).
On the Unseen Route, PPO Fairness demonstrates superior generalisation, retaining 130.7%
of its training throughput (459 → 600 vehicles) while maintaining a Jain’s Index of 0.8314,
well above the 0.8 acceptability threshold. Notably, all fairness-weighted controllers maintain acceptable directional equity under distribution shift.
A key finding is that fairness-oriented reward shaping implicitly regularises the learned
policy, improving robustness under distribution shift. These results offer practical deployment guidelines for smart city traffic management systems.
One-Dimensional Thermal Modeling of Falling-Particle Solar Receivers
(university of ghardaia, 2026) DAOUDI Bakir; REHAIEM Nour elhouda
he change to temperature concentrated solar power needs good receiver designs. This
work shows a model of a free falling particle receiver. The model is made up of 40 parts and it
looks at how heat moves through radiation, convection and the movement of falling particles.
When we compare this model to data we see that it is very accurate with a small error of only
1.18% for the temperature of the particles. We also did a study to see what factors are most
important. We found that how well the particles absorb heat is the most important thing,
followed by the amount of heat and the rate at which the particles move.
Our analysis shows that we can get high thermal efficiencies, up to 84.5% when we use
a lot of heat. We found some things, like how wind outside can change when the system starts
to work from 0.5 to 0.8 megawatts per square meter. We also found that when the particles are
around 500 micrometers in size they start to behave.The study also shows that there is a
balance to be found where the best performance comes from using particles rather than
making them too hot. Finally we made some maps that show how efficient the system can be
when we use amounts of particles and this can help us design better systems using computers
and optimization techniques with a fixed temperature of 1073.15 Kelvin. Concentrated solar
power systems, like these can be very useful and high temperature concentrated solar power
systems need designs so we need to keep working on concentrated solar power and high
temperature concentrated solar power systems
Etude et commande d’un convertisseur élévateur a trois niveaux intégrés dans un système PV-Batterie
(university of ghardaia, 2026) RECIOUI NOUR EL HOUDA
Ce travail porte sur l’étude, la modélisation et la commande d’un système multi-sources
PV/Batterie intégrant un convertisseur élévateur à trois niveaux. L’objectif principal est
d’assurer une gestion optimale de l’énergie produite par les panneaux photovoltaïques et
stockée dans la batterie tout en garantissant une tension stable au niveau du bus continu
(DC-Link) et une alimentation fiable de la charge.
Le convertisseur hacheur élévateur à trois niveaux constitue une amélioration du convertisseur
conventionnel à deux niveaux grâce à la réduction des pertes de commutation, à l’amélioration
du rendement énergétique et à la diminution des ondulations de tension et de courant. Une
stratégie de commande basée sur un régulateur PI associée à un algorithme MPPT a été
développée afin d’extraire la puissance maximale disponible du générateur photovoltaïque et de
maintenir la stabilité du bus continu.
Les simulations réalisées sous MATLAB/Simulink ont permis d’évaluer le comportement
dynamique du système dans différentes conditions d’irradiance et de charge. Les résultats
obtenus démontrent l’efficacité de la structure proposée et mettent en évidence la supériorité
du convertisseur élévateur à trois niveaux par rapport au convertisseur classique à deux
niveaux en termes de stabilité de tension, de qualité de l’énergie et de performances globales.
Production et caractérisation d’un adsorbant biosourcé à partir des déchets de palmier dattier
(university of ghardaia, 2026) Bahaz Amna; Rouighi Nivine Maroua
Cette étude a pour objectif l’élaboration d’un biosorbant à faible coût, obtenu à
partir de pétiole du palmier dattier, destiné à l’élimination du colorant organique orange
de méthyle (OM) présent dans les eaux contaminées, par un procédé d’adsorption. Le
matériau adsorbant (PT) a été chimiquement activé à l’aide d’hydroxyde de potassium
(KOH), puis caractérisé par la détermination du point de charge nulle (pHPZC), évalué
9,5.
L’étude expérimentale a porté sur l’influence de plusieurs paramètres
opératoires, notamment le pH, la température, le temps de contact et la concentration
initiale du colorant, afin d’évaluer les performances adsorption du matériau préparé.
Les résultats obtenus montrent une capacité d’adsorption élevée du biosorbant vis-à-vis
de l’orange de méthyle.
L’analyse des données expérimentales révèle un bon ajustement aux modèle
cinétique de pseudo-second ordre, ainsi qu’aux modèle d’isotherme de Langmuir. La
capacité maximale d’adsorption (Qm) déterminée à partir du modèle de Langmuir a été
estimée à 51,8 mg·g−1. Par ailleurs, l’étude thermodynamique indique que le processus
d’adsorption est spontané et endothermique.
À la lumière de ces résultats, le biosorbant élaboré apparaît comme une
alternative prometteuse, économique et efficace pour l’élimination des polluants
organiques dans les eaux usées, avec un potentiel d’application dans les technologies
de traitement des eaux.
Speech Denoising Using Diffusion Model Technique
(university of ghardaia, 2026) Laouar Meriem; Zahouani Sarra
Speech is the primary medium for human communication, yet it is almost
always degraded by background noise in real-world environments. This noise
reduces intelligibility and harms the performance of critical applications such as
automatic speech recognition, telecommunications, and audio streaming. Speech
denoising, therefore, plays a vital role in restoring clean speech and ensuring
reliable human-machine interaction.
This thesis explores diffusion-based generative models for speech denoising,
focusing on the DiffWave architecture. The objective is to build a system that
produces high-quality enhanced speech while operating fast enough for real-world
deployment. To this end, two diffusion step configurations are compared: T=50
and T=200. Although T=200 provides a finer discretization, T=50 achieves better
generalization on unseen data (SI-SDR: 10.93 dB vs 9.94 dB), with an optimal
inference step of t=5. This is primarily because T=50 was trained for significantly
longer (240,000 iterations compared to 65,000 for T=200) due to GPU constraints,
allowing it to converge more effectively.
To overcome inference latency, Denoising Diffusion Implicit Models (DDIM)
are integrated, reducing sampling steps from 50 to just 3. This achieves a real-
time factor (RTF) of 0.0732, corresponding to 13.7× real-time performance, with
a minimal quality loss of only -0.64 dB.
A key contribution is the evaluation on Arabic speech. Although trained
exclusively on English digits, the model achieves its best performance on the
Arabic dataset (SI-SDR = 3.94 dB), with the majority of test samples showing
improvement after processing, confirming strong cross-lingual generalization. This
is attributed to the model’s ability to learn latent acoustic representations that
transcend language boundaries.
The model was also evaluated on French speech using the Common Voice
dataset, demonstrating clear cross-lingual generalization capabilities with promis-
ing results that support its effectiveness in diverse linguistic environments.
The model is also validated on real-world recordings from office, street, and
cafeteria environments, confirming its practical applicability beyond controlled
laboratory conditions. These results demonstrate that the proposed approach
offers a practical, fast, and language-robust solution for speech denoising, suitable
for real-time applications.