Archive ouverte de l'ENAC
Not a member yet
3458 research outputs found
Sort by
Scalable Syndrome-based Neural Decoders for Bit-Interleaved Coded Modulations
No full textInternational audienceIn this work, we introduce a framework that enablesthe use of Syndrome-Based Neural Decoders (SBND) for high-order Bit-Interleaved Coded Modulations (BICM). To this end,we extend the previous results on SBND, for which the validityis limited to Binary Phase-Shift Keying (BPSK), by means of atheoretical channel modeling of the bit Log-Likelihood Ratio (bit-LLR) induced outputs. We implement the proposed SBND systemfor two polar codes (64, 32) and (128, 64), using a RecurrentNeural Network (RNN) and a Transformer-based architecture.Both implementations are compared in Bit Error Rate (BER)performance and computational complexity.<br /
Energy Efficient Matrix Computations Through Homomorphic Compression
No full textInternational audienceScientific computing is one of the most energy-intensive areas of computer science. Petabytes of data are processed and stored for a single experiment while super-computers consume tens of megawatts. These last years, new lossy compression techniques dedicated to scientific data have been introduced, such as zfp and sz. They allow one to dras-tically reduce the size of the data but they require additional computations for compression and decompression. Recently, a new homomorphic compressor, named blaz, has been developed which makes it possible to perform matrix operations directly among the compression data. Important gains are obtained since decompression and re-compression operations are avoided and since less operations are needed to compute among the compressed data. In this article, we show that using blazfor linear algebra operations among matrices may reduce the energy consumption by a factor 10 compared to standard operations (without any compression) and by a factor 100 compared to the zfp. Beside measures on sequences of matrix operations, a case study in data analysis in presented
A comparative study of gravity models to assess the evolution of urban bicycle mobility
No full textInternational audienc
Double task switching: An investigation into the effects of similarity and task-rule congruency on cognitive flexibility
No full textInternational audienceSimilarity between tasks is an understudied factor in research on cognitive flexibility. This behavioural experiment had 31 participants perform a task switch paradigm in which participants were required to switch between 4 tasks of varying similarity. The experiment was constructed in a way that simultaneously allows for investigating the impact of mental fatigue and task-rule congruency on the participants. The results indicate that similarity between tasks substantially impacts performance with different effects on RT and accuracy. While learning effects may have negated the impact of mental fatigue across the 5 experimental blocks, a significant decrease in performance was observed within blocks. Furthermore, the exploratory analysis proposes a novel interaction between task-rule incongruent trials and the task of the previous trial. These results support the notion that neither the interference view of cognitive flexibility nor the reconfiguration view are fully adequate at explaining task switch costs if similarity is added as a factor. The presented study presents strong evidence that fundamental findings in the domain of cognitive flexibility may not map linearly to more ecological settings where tasks are often more dissimilar
Commande à Modèle Restreint appliquée au Quanser Aéro avec des paramètres incertains
No full textInternational audienceThis work study the impact on a Model-Free Control (MFC) feedback combined with flatness-based control, also know as Restricted-Model Control (RMC), in order to address the problem of robustness to uncertain parameters. The proposed control architecture, is applied to the half-quad Quanser Aero benchmark in two degree-of-freedom configuration. Tests focuse on the robustness of RMC to over-efficient propellers, under-efficient propellers and augmented friction
A survey on reinforcement learning in aviation applications
No full textInternational audienceReinforcement learning (RL) has emerged as a powerful tool for addressing complex decision making problems in various domains, including aviation. This paper provides a comprehensive overview of RL and its applications in the aviation industry. We begin by introducing the fundamental concepts and algorithms of RL, highlighting their unique advantages in learning from interaction and optimizing decision-making processes. We then delve into a detailed examination of the successful implementation of RL methods in aviation, covering areas such as flight control, air traffic management, airline revenue management, aircraft maintenance scheduling, etc. Furthermore, we discuss the potential benefits of RL in enhancing safety, and sustainability within the aviation sector. Finally, we identify and explore open challenges and areas for future research, emphasizing the need for continued innovation and collaboration between the fields of reinforcement learning and aviation
Application of Reduced-Order Robust Control to Multi-Rotor Stabilization and Guidance
No full textInternational audienceThis paper presents a cascaded robust control scheme for fully actuated hexacopter drones. The system model is first derived using Newton-Euler equations and subsequently linearized. H∞ synthesis is then used to design controllers for thrust, stabilization and guidance dynamics. A structured constrained formulation is used to address the problem, allowing for a significant reduction in the controllers’ order compared to the classical full-order approach, while keeping the same robustness and performance levels. The effectiveness of the complete control architecture is assessed using linear tools, and a complete MATLAB simulator involving the full nonlinear model of the drone is implemented. Two main operational scenarios are considered: hovering with a cable-suspended pendulum and flying through predefined waypoints, each time in the presence of external perturbations
Stochastic Modeling of the Radiation Pattern of an Airborne Single-Band GNSS Antenna
No full textInternational audienceAirborne antennas play a critical role in aircraft systems, serving as channels for signals entering and exiting the aircraft. As their presence continues to grow, ensuring electromagnetic compatibility and optimal performance across various environments becomes increasingly imperative. However, traditional methods of characterizing these antennas through measurements or simulations encounter challenges due to logistical constraints and computational limitations. To address these issues, this paper proposes a theoretical approach to characterize the stochastic radiation pattern of an airborne L1 GNSS patch antenna. This approach considers randomness in both the antenna parameters and the aircraft geometry. Invoking a two-step methodology involving Vector Spherical Harmonics (VSH) and Polynomial Chaos (PC) expansions, the paper establishes a framework to analyze the spatial distribution and stochastic nature of the radiated electric field. The PC expansion provides easy and efficient computing of statistics such as mean, variance, and sensitivity indices, outperforming classical Monte Carlo schemes. Applying this methodology to the airborne L1 GNSS patch antenna results in a stochastic radiation pattern, with identified boundaries encapsulating most field realizations with high confidence. Simplified aircraft geometry, reduced to the fuselage, enhances computational tractability in the L1 GNSS patch antenna case. By offering a systematic approach to characterizing airborne antenna radiation patterns, this paper contributes to bypassing systematic measurement or simulation for each aircraft-antenna pair
Étude de la dispersion de cristaux diélectriques imprimés en 3D pour des applications d'antennes à résonateur diélectrique
No full textInternational audienceLa conception d’antennes à résonateur diélectrique (ARD) imprimées en 3D avec des cellules unitaires périodiques requiert la connaissance de la permittivité effective du diélectrique artificiel. Dans ce but, troisméthodes permettant d’extraire cette permittivité effective sont comparées. Nous montrons également qu’il estpossible de concevoir des antennes à résonateur diélectrique composées de cellules unitaires en régime dispersifà l’aide d’une de ces méthodes, la méthode d’expansionde l’onde plane. Ce résultat ouvre la possibilité de concevoir des ARD imprimées en 3D fonctionnant à plus hautesfréquences que les solutions actuelles
Optimisation des opérations aéroportuaires en cas de perturbation des modes d'accès afin d'améliorer l'expérience passager
No full textIn light of the climate crisis and air transportation regulations, there is a growing interest in using public transportation or trains to replace cars and short-haul flights to access hub airports. In order to ensure passengers with reliable door-to-door journeys, seamless integration between air and ground transport modes is required. This integration is essential, especially during disruptions such as train or subway shutdowns, which significantly increase the risk of passengers arriving late and missing their flights. As part of the European TRANSIT project, dedicated to developing coordination mechanisms between air and ground transport operations based on information sharing, we propose solutions that can be implemented at the airport level to deal with airport access mode disruptions. Assuming real-time information sharing between air and ground transport stakeholders, our approach involves tactical adjustments in airport operations to enhance the overall passenger experience. Our work presents different contributions. Firstly, we propose a comprehensive framework for airport and passenger flow modelling. We also present the finding of a suitable parametric distribution for modelling the passenger arrival profile of a flight based on its characteristics, allowing for a reconstruction of passenger arrival flow at the airport entrance based on a flight schedule. Secondly, we propose two recovery strategies for airport operators to mitigate the impact of disruptions on passengers. Both strategies are grounded in Operations Research, employing mathematical modelling and algorithmic solving approaches to help decision-making. The first strategy is a tactical reallocation of security teams to reduce passenger waiting times at security screening checkpoints. We propose two problem formulations: whether or not to consider passenger flow segregation to handle disrupted passengers in a dedicated fast-track lane. The second recovery strategy is a tactical flight rescheduling to wait for delayed passengers while considering airside operational constraints and air-connecting passengers. We propose different approaches for solving the related optimization problems, namely ILP-solving, simulated annealing and greedy heuristic approaches and compare their effectiveness. We also propose a sliding-time window approach to implement such recovery strategies when dealing with operational aspects and data availability. We test both recovery strategies on a benchmark of disruption scenarios to gain insights and evaluate the benefits of such disruption management solutions for passengers. Finally, we extend the research beyond theoretical models to practical applications. First, we combine the security team reallocation strategy with discrete simulation tools to provide realistic assessments and the basis of a decision support system for airport operators. Secondly, we broaden the scope of flight rescheduling to the air-rail network level, capturing delay propagation and its impact on the overall transportation network. This extension allows for a comprehensive evaluation of passenger delays at their final destinations and an assessment of the benefits obtained for passengers through a tactical flight rescheduling adjustment.Dans un contexte de crise climatique et de régulations du transport aérien, les transports publics ou les trains sont amenés à remplacer les véhicules privés et les vols court-courriers pour accéder aux aéroports. Afin de garantir aux passagers des trajets porte-à-porte fiables, il est alors nécessaire d'assurer une coordination entre les modes de transport aérien et terrestre. Cette intégration est essentielle notamment en cas de perturbations, telles qu'une panne de métro ou de train, induisant un risque pour les passagers d'arriver en retard à l'aéroport et éventuellement de manquer leur vol. Dans le cadre du projet européen TRANSIT, dédié au développement de mécanismes de coordination entre acteurs du transport aérien et terrestre basés sur le partage d'informations, nous proposons dans cette thèse des solutions pouvant être mises en œuvre au niveau de l'aéroport pour limiter l'impact de perturbations sur l'accès de l'aéroport pour les passagers. Supposant un échange d'informations en temps réel entre les opérateurs de transport aérien et terrestre, nous proposons des ajustements tactiques des opérations aéroportuaires afin d'améliorer l'expérience globale du passager. Cette thèse présente différentes contributions. Tout d'abord, nous avons développé un cadre de modélisation d'un aéroport et de son flux de passagers associé. Nous y présentons également une distribution de probabilité paramétrique appropriée pour modéliser le profil d'arrivée des passagers d'un vol en fonction de ses caractéristiques, permettant de reconstruire le flux d'arrivée des passagers à un aéroport à partir d'un planning de vol. Dans un deuxième temps, nous proposons deux stratégies pour les opérateurs aéroportuaires afin d'atténuer l'impact des perturbations sur les passagers. Ces deux stratégies sont basées sur des méthodes de Recherche Opérationnelle, faisant appel à des modèles mathématiques et des résolutions algorithmiques pour aider à la prise de décision. La première stratégie consiste en une réaffectation tactique des équipes de sécurité afin de réduire les temps d'attente des passagers aux points de contrôles de sûreté. Nous proposons deux formulations du problème, incluant ou non une ségrégation des flux de passagers pour traiter les passagers impactés par la perturbation dans une file d'attente séparée. La seconde stratégie consiste en une replanification tactique des vols pour attendre les passagers retardés tout en tenant compte des contraintes opérationnelles côté piste et des passagers en correspondance. Plusieurs approches de résolutions sont mises en place et comparées pour résoudre les différents problèmes d'optimisation. Ces méthodes reposent sur de la PLNE, un recuit simulé ou encore des heuristiques gloutonnes. Nous proposons également une approche par fenêtre de temps glissante pour mettre en œuvre ces stratégies en tenant compte des aspects opérationnels et de la disponibilité des données passagers. Nous testons les deux stratégies sur différents scénarios de perturbation afin de mieux comprendre et d'évaluer les bénéfices de ces solutions pour les passagers. Enfin, nous étendons la recherche au-delà des modèles théoriques à des applications pratiques. Premièrement, nous proposons une intégration de la stratégie de réaffectation des équipes de sûreté avec de la simulation à événement discret pour fournir une évaluation plus réaliste et établir la base d'un système d'aide à la décision pour les opérateurs aéroportuaires. Nous élargissons également le champ d'application de la replanification des vols au niveau du réseau air-rail, permettant une prise en compte de la propagation des retards et de son impact sur l'ensemble du réseau de transport. Cette extension permet une évaluation des retards des passagers à leur destination finale et l'évaluation de la pertinence d'une replanification tactique des vols pour les passagers