Дисертації з теми "Intelligence artificielle – Consommation d'énergie"
Оформте джерело за APA, MLA, Chicago, Harvard та іншими стилями
Ознайомтеся з топ-23 дисертацій для дослідження на тему "Intelligence artificielle – Consommation d'énergie".
Біля кожної праці в переліку літератури доступна кнопка «Додати до бібліографії». Скористайтеся нею – і ми автоматично оформимо бібліографічне посилання на обрану працю в потрібному вам стилі цитування: APA, MLA, «Гарвард», «Чикаго», «Ванкувер» тощо.
Також ви можете завантажити повний текст наукової публікації у форматі «.pdf» та прочитати онлайн анотацію до роботи, якщо відповідні параметри наявні в метаданих.
Переглядайте дисертації для різних дисциплін та оформлюйте правильно вашу бібліографію.
Ihsane, Imane. "Prévision à court terme et gestion des consommations d'énergie électrique dans l'habitat." Thesis, Nantes, 2020. http://www.theses.fr/2020NANT4019.
In this thesis, a short-term forecasting model of electricity demand based on artificial neural networks was developed. First of all, particular care was taken in the selection of the relevant input variables of this model. Then it was used to reproduce the load curves at the scale of an individual housing and at scale of a region. The comparaison between these two levels of aggregation highlighted the limitations of the indicators for assessing the quality of the forecast. As a result, new indicators adapted to residential load profiles were created. In particular for error detection during peak periods. The thesis work also presents an approach to managing the demand for electrical loads. The particularity of this strategy lies in the consumer’s participation in reducing peak electricity consumption and in benefiting from advantageous off-peak electricity rates, which makes him an active consumer. Emphasis will also be placed on consumer comfort. The principle consists in establishing an allocation of priorities to electrical loads. Depending on these and on a setpoint power to be respected, the algorithm grants a satisfaction (or not) to the requests for activation of the loads. In the absence of sufficient usable power, unmet demand, in particular from heating systems, can impact the thermal comfort of the user. In order to remedy this problem, the management methodology has been improved by combining it with the short-term forecasting of load consumption. This forecast identifies the heaters whose activation will be interrupted in the future and anticipates their activation under power and priority constraints. The results obtained are promising and validate the effectiveness of the proposed methodologies
Chaib, Draa Ismat Yahia. "Optimisation de la consommation d'énergie des systèmes mobiles par l'analyse des besoins de l'utilisateur." Thesis, Valenciennes, 2018. http://www.theses.fr/2018VALE0030/document.
Optimizing energy consumption in modern mobile handled devices plays a crucial role as lowering the power consumption impacts battery life and systemreliability. Recent mobile platforms have an increasing number of sensors and processing components. Added to the popularity of power-hungry applications, battery life in mobile devices is an important issue. However, the utilization pattern of large amount of data from the various sensors can be beneficial to detect the changing device context, the user needs and the running application requirements in terms of hardware resources. When these information are used properly, an efficient control of power knobs can be implemented to reduce the energy consumption. This thesis has been achieved in collaboration with Intel Portland
Paudel, Subodh. "Methodology to estimate building energy consumption using artificial intelligence." Thesis, Nantes, Ecole des Mines, 2016. http://www.theses.fr/2016EMNA0237/document.
High-energy efficiency building standards (as Low energy building LEB) to improve building consumption have drawn significant attention. Building standards is basically focused on improving thermal performance of envelope and high heat capacity thus creating a higher thermal inertia. However, LEB concept introduces alarge time constant as well as large heat capacity resulting in a slower rate of heat transfer between interior of building and outdoor environment. Therefore, it is challenging to estimate and predict thermal energy demand for such LEBs. This work focuses on artificial intelligence (AI) models to predict energy consumptionof LEBs. We consider two kinds of AI modeling approaches: “all data” and “relevant data”. The “all data” uses all available data and “relevant data” uses a small representative day dataset and addresses the complexity of building non-linear dynamics by introducing past day climatic impacts behavior. This extraction is based on either simple physical understanding: Heating Degree Day (HDD), modified HDD or pattern recognition methods: Frechet Distance and Dynamic Time Warping (DTW). Four AI techniques have been considered: Artificial Neural Network (ANN), Support Vector Machine (SVM), Boosted Ensemble Decision Tree (BEDT) and Random forest (RF). In a first part, numerical simulations for six buildings (heat demand in the range [25 – 85 kWh/m².yr]) have been performed. The approach “relevant data” with (DTW, SVM) shows the best results. Real data of the building “Ecole des Mines de Nantes” proves the approach is still relevant
Ouni, Bassem. "Caractérisation, modélisation et estimation de la consommation d'énergie à haut-niveau des OS embarqués." Phd thesis, Université Nice Sophia Antipolis, 2013. http://tel.archives-ouvertes.fr/tel-01059814.
Rioual, Yohann. "RL-based Energy Management for Autonomous Cyber Physical Systems." Thesis, Lorient, 2019. http://www.theses.fr/2019LORIS533.
The energy management of a cyber physical system is a difficult task because of the complexity of hardware architectures and the use of OS. In addition, these systems are deployed in changing environments where their energy harvesting capacity varies. Over time, their energy consumption is modified due to the ageing of the components. Consumption models designed in the laboratory cannot take into account all possible deployment scenarios and system aging. One approach that is developing is the use of reinforcement learning in which we no longer know the system's consumption model; but thanks to this approach, the system is still able to adapt its operation during its deployment according to the evolution of its environment. Several approaches exist in reinforcement learning. The first part of this thesis is devoted to proposing guidelines to help for selecting the most appropriate approach for a given application and target. The second part of this thesis focuses on the design of the reward we give to our system that will influence its behaviour in its environment. Two reward functions able to adjust the system’s performance according to the energy available are presented. The third and last part of this thesis explores the use of several agents to independently control the different modules of our system. This approach allows a more precise control of energy consumption, reducing memory usage compared to a single agent approach
Cherdo, Yann. "Détection d'anomalie non supervisée sur les séries temporelle à faible coût énergétique utilisant les SNNs." Electronic Thesis or Diss., Université Côte d'Azur, 2024. http://www.theses.fr/2024COAZ4018.
In the context of the predictive maintenance of the car manufacturer Renault, this thesis aims at providing low-power solutions for unsupervised anomaly detection on time-series. With the recent evolution of cars, more and more data are produced and need to be processed by machine learning algorithms. This processing can be performed in the cloud or directly at the edge inside the car. In such a case, network bandwidth, cloud services costs, data privacy management and data loss can be saved. Embedding a machine learning model inside a car is challenging as it requires frugal models due to memory and processing constraints. To this aim, we study the usage of spiking neural networks (SNNs) for anomaly detection, prediction and classification on time-series. SNNs models' performance and energy costs are evaluated in an edge scenario using generic hardware models that consider all calculation and memory costs. To leverage as much as possible the sparsity of SNNs, we propose a model with trainable sparse connections that consumes half the energy compared to its non-sparse version. This model is evaluated on anomaly detection public benchmarks, a real use-case of anomaly detection from Renault Alpine cars, weather forecasts and the google speech command dataset. We also compare its performance with other existing SNN and non-spiking models. We conclude that, for some use-cases, spiking models can provide state-of-the-art performance while consuming 2 to 8 times less energy. Yet, further studies should be undertaken to evaluate these models once embedded in a car. Inspired by neuroscience, we argue that other bio-inspired properties such as attention, sparsity, hierarchy or neural assemblies dynamics could be exploited to even get better energy efficiency and performance with spiking models. Finally, we end this thesis with an essay dealing with cognitive neuroscience, philosophy and artificial intelligence. Diving into conceptual difficulties linked to consciousness and considering the deterministic mechanisms of memory, we argue that consciousness and the self could be constitutively independent from memory. The aim of this essay is to question the nature of humans by contrast with the ones of machines and AI
Herzog, Christina. "Contributions à la modélisation avec un système multi agent du transfert technologique en Green IT." Thesis, Toulouse 3, 2015. http://www.theses.fr/2015TOU30178/document.
Over the past 5 to 10 years, research is numerous on energy reduction in IT (mainly electricity reduction). Several studies indeed alerted the stakeholders and environmental agencies on the urgency of the problem of the energy consumption of large scale infrastructures, like data centres, clouds or simply companies running servers and lots of IT equipment. This awareness moved from a non-so-important issue to major constraints on the operation of these infrastructures. In some cases, the operational costs reach the investment costs, urging new methodologies to appear in order to reduce costs and ecological impact. As of today, new hardware are developed by equipment manufacturers to decrease these costs. Only few and basic techniques are offered at the software and middleware levels out-of-the-box: But in laboratories, some techniques have proven on synthetic data, dedicated workflows or selected applications, to be able to save energy during the lifetime of an infrastructure, in several contexts, from Cloud to HPC in particular. Unfortunately, the transfer (or even the knowledge of the existence) of these techniques to industries is limited to project partners, innovative companies or large private research centres, able to invest time (thus money) on this topic. In my thesis, I investigate the reasons restraining the large adoption of several research results, from the simpler ones to more elaborated ones and I model the ties and interactions between the actors of the technological transfer. The target field has been restricted to Green IT but the methodology and the developed models can be extended to other domains as well. The idea is to identify, on the scale of technical maturity for wider adoption, what is missing and how to increase the speed of the transfer of scientific knowledge. The methodology is based on the following path: First, identifying the actors involved in the process of technology transfer, and understanding their motivations, their means of actions and their limitations. After a study of the state of the art in the domain of innovation diffusion and innovation management, this phase involved the production and the analysis of a dedicated survey targeting researchers and companies, from different size and turnover, restricted to those working in the Green IT field. Identifying each actor is not sufficient since they all interact; therefore their links and the potential of these links for technology transfer have also been studied carefully in a second phase so as to identify the most important ones, with the same methodology with the actors' identification. From these two phases, a multi-agent system (MAS) has been designed
Lesel, Jonathan. "Optimisation de la consommation énergétique d'une ligne de métro automatique prenant en compte les aléas de trafic à l'aide d'outils d'intelligence artificielle." Thesis, Paris, ENSAM, 2016. http://www.theses.fr/2016ENAM0018/document.
In 2014, as part of the Climate Plan, EU member countries have committed to reduce by 27% their energy consumption. One of the main focal areas consists in increasing the energy efficiency of urban transports. This thesis aims to propose a methodology to reduce the energy consumption of automatic metro lines while integrating traffic disruptions that occur under normal operating conditions. The principle adopted in this work is to maximize the reuse of electrical energy generated during braking of the train, by other trains running on the line. First part is dedicated to the electrical modeling of an automatic metro line and development of methods to calculate power flows between trains and power substations. Then, optimization algorithms are introduced to perform optimization of the most influential operating parameters in an ideal configuration ignoring traffic fluctuations. Finally, a methodology based on learning simulation data is developed in order to achieve optimization of energy consumption integrating traffic disruptions in real time. This last part will thus purchase the objective to provide a decision support to determine optimal dwell times to be carried out by trains in each station, so as to maximize braking energy recovery
Ali, Sadaqat. "Energy management of multi-source DC microgrid systems for residential applications." Electronic Thesis or Diss., Université de Lorraine, 2023. http://www.theses.fr/2023LORR0159.
Compared to the alternating current (AC) electrical grid, the direct current (DC) electrical grid has demonstrated numerous advantages, such as its natural interface with renewable energy sources (RES), energy storage systems, and DC loads. It offers superior efficiency with fewer conversion steps, simpler control without skin effect or reactive power considerations. DC microgrids remain a relatively new technology, and their network architectures, control strategies, and stabilization techniques require significant research efforts. In this context, this thesis focuses on energy management issues in a multi-source DC electrical grid dedicated to residential applications. The DC electrical grid consists of distributed generators (solar panels), a hybrid energy storage system (HESS) with batteries and a supercapacitor (SC), and DC loads interconnected via DC/DC power converters. The primary objective of this research is to develop an advanced energy management strategy (EMS) to enhance the operational efficiency of the system while improving its reliability and sustainability. A hierarchical simulation platform of the DC electrical grid has been developed using MATLAB/Simulink. It comprises two layers with different time scales: a local control layer (time scale of a few seconds to minutes due to converter switching behavior) for controlling local components, and a system-level control layer (time scale of a few days to months with accelerated testing) for long-term validation and performance evaluation of the EMS. In the local control layer, solar panels, batteries, and the supercapacitor have been modeled and controlled separately. Various control modes, such as current control, voltage control, and maximum power point tracking (MPPT), have been implemented. A low-pass filter (LPF) has been applied to divide the total HESS power into low and high frequencies for the batteries and supercapacitor. Different LPF cutoff frequencies for power sharing have also been studied. A combined hybrid bi-level EMS and automatic sizing have been proposed and validated. It mainly covers five operational scenarios, including solar panel production reduction, load reduction, and three scenarios involving HESS control combined with supercapacitor state of charge (SOC) control retention. An objective function that considers both capital expenditure (CAPEX) and operating costs (OPEX) has been designed for EMS performance evaluation. The interaction between the HESS and EMS has been jointly studied based on an open dataset of residential electrical consumption profiles covering both summer and winter seasons. Finally, an experimental platform of a multi-source DC electrical grid has been developed to validate the EMS in real-time. It comprises four lithium-ion batteries, a supercapacitor, a programmable DC power supply, a programmable DC load, corresponding DC/DC converters, and a real-time controller (dSPACE/Microlabbox). Accelerated tests have been conducted to verify the proposed EMS in different operational scenarios by integrating real solar panels and load consumption profiles. The hierarchical simulation and experimental DC electrical grid platforms can be generally used to verify and evaluate various EMS
Bouabdallaoui, Yassine. "Introduction de l'intelligence artificielle dans le secteur de la construction : études de cas du Facility Management." Electronic Thesis or Diss., Centrale Lille Institut, 2021. http://www.theses.fr/2021CLIL0022.
The industry of Facility Management (FM) has known a rapid advancement through the last decades which leads to a largeexpansion of FM activities. The FM organisations have evolved from the traditional role of providing maintenance services toinclude complex and interconnected activities involving people, processes and technologies. As a consequence of thisexponential growth, facility managers are dealing with growing and varied challenges ranging from energy efficiency andenvironmental challenges to service customisation and customer satisfaction. The development of Artificial Intelligence (AI)is offering academics and practitioners a new set of tools to address these challenges. AI is enabling multiple solutions suchas automation, improving predictability and forecasting and offering services customisation. The Facility Managementindustry can benefit from these new techniques to better manage their assets and improve their processes. However, theintegration of AI into the FM ecosystem is a challenging task that needs to overcome the gap between the business driversand the AI. To unlock the full potential of data analytics and AI in the FM industry, significant work is needed to overcomethe issues of data quality and data management in the FM sector. The overall aim of this thesis is to conceptualise thetheoretical and practical understanding and implementation of artificial intelligence and data-driven technologies into FacilityManagement activities to leverage data and optimise facilities usage. Promises of AI implementations were presented alongwith the challenges and the barriers limiting the development of AI in the FM sector. To resolve these issues, a frameworkwas proposed to improve data management and leverage AI in FM. Multiple case studies were selected to address thisframework. The selected case studies covered predictive maintenance, virtual assistant and natural language processingapplications. The results of this work demonstrated the potential of AI to address FM challenges such in maintenancemanagement and waste management. However, multiple barriers limiting the development of AI in the FM sector wereidentified including data availability issues
Bonte, Mathieu. "Influence du comportement de l'occupant sur la performance énergétique du bâtiment : modélisation par intelligence artificielle et mesures in situ." Toulouse 3, 2014. http://thesesups.ups-tlse.fr/2495/.
Building sector plays a major role in global warming. In France, it is responsible of about 40% of energy consumption et about 33% of carbon emissions. In this context, building designers try to improve building energy performance. To do so, they often use building energy modeling (BEM) software to predict future energy use. For several years now, researchers have observed a difference between actual and predicted energy performance. Some reasons are pointed out such as uncertainties on physical properties of building materials and lack of precision of fluid dynamics models. One of the main causes could come from bad assessments in the modeling of occupant behavior. Occupant is often considered as passive in building simulation hypothesis. However, numerous of papers show that he act on the building he is in, and on personal characteristics. The work presented here intend to characterize occupant behavior and its influence on energy use. In the first part of the manuscript we assess the individual impact of several actions using design of experiments (DOE) methodology. Actions like operations on windows, blind or thermostat are investigated separately. We show that two opposite extreme behaviors (economic and wasteful) could lead to significant difference in building energy use. Moreover, a factor two-to-one in total energy use is observed between passive and active behaviors. In the second part we focused on an experimental approach. Thermal and visual environment of 4 offices have been monitored during a year and online questionnaires about comfort and behavior have been submitted to office occupants. Tank to a statistical analysis we estimates probabilities of acting on windows, blinds and clothing insulation against physical variables or thermal sensation. Final part of the thesis deals with the development of an occupant behavior model called OASys (Occupant Actions System) and running under TRNSys software. The model is based on an artificial intelligence algorithm and is intended to predict occupant interactions with thermostat, clothing insulation, windows, blinds and lighting system based on thermal and visual sensation. Results from OASys are compared to results from literature through various case studies for partial validation. They also confirm the significant impact of occupant behavior on building energy performance
Sardouk, Ahmad. "Agrégation de données dans les réseaux de capteurs sans fil à base d'agents coopératifs." Troyes, 2010. http://www.theses.fr/2010TROY0013.
The main role of Wireless Sensor Network is to collect information from the environment by a high number of Sensor Nodes (SNs). The SNs have a lifetime limited by their batteries. Hence, SNs that ran out of battery will be out of the network and may create serious network partitioning and information loss problems. Thus, in order to maximize the WSN lifetime, it is important to minimize the power consumption of each SN and better manage the consumption of nodes that are in critical positions of the network. As the radio communication is the main power consumer, we propose a multi-agent based data aggregation solution, which reduces the amount of communicated information and hence reduces the power consumption of the SNs. We propose to implement in each node an agent that manages optimally the SN, processes locally its information and estimates their importance. The implemented agents cooperate together to eliminate the inter-SN redundancy and the useless information and to create a message summarizing the network’s important information. The agent manages the power consumption of each node according to its position in the network, the nodes density in its coverage zone, its residual battery and the importance of its current information. This management aims to balance the power consumption of the SNs and to maximize the life-time of SNs in critical positions to avoid the network partitioning
Sanoussi, Hamadou. "Énergie et économie : analyse de la relation consommation d'électricité et production de richesse dans une perspective d'intelligence économique." Thesis, Lyon 3, 2014. http://www.theses.fr/2014LYO30004.
The subject of this thesis consists of an analysis of the relationship between electricity consumption and Gross Domestic Product from the perspective of Competitive Intelligence. More specifically, it analyzes the evolution of the electrical intensity of economic activity from 2003 to 2012 in the developed countries of the G7, and then estimates their electricity needs from 2013 to 2022. Part one attempt to explore theoretical and practical aspects of Competitive Intelligence to understand and apply them, while part two is devoted to the empirical analysis itself.Concerning the latter, our results are as follows:First, the electrical intensity curves of two countries—Canada and the United States—dominate those of other developed countries; thus, the economies of these two North American countries are more energy-hungry than those of Japan and the countries of the European Union. The overall temporal evolution of electricity consumption per GDP unit over a ten-year period (2003-1012) has gone down in five countries: Canada (-12%), the United Kingdom (-5.3%), the United States (-5%), France (-4%), and Germany (-3%). On the other hand, this evolution has gone the other direction in Japan (+5%) and Italy (+6%). The effect of “structure” is negative across all analyzed data, suggesting general “tertiarisation”. However, the effect of “electricity efficiency” is mixed: it is negative in the United States and Canada, but positive for the rest of group.Second, estimations indicate an overall growth in electricity demand across all G7 countries from 2013 to 2022. Additionally, electrical elasticity coefficients/GDP units are down in all countries except Italy. This tells us that the average annual demand for electricity in these countries should increase at a slower rate than their respective GDPs.Lastly, the primary research perspectives that appear at the beginning of this thesis concern the transposition of our model of analysis (energetic intelligence) onto other forms of energy such as oil, natural gas, coal, and renewable energy sources. In the end, this model could be useful to economic and political authorities (governments, private companies, NGOs, IGOs, etc.) as an instrument of economic, energy, and environmental policy
Bedecarrats, Thomas. "Etude et intégration d’un circuit analogique, basse consommation et à faible surface d'empreinte, de neurone impulsionnel basé sur l’utilisation du BIMOS en technologie 28 nm FD-SOI." Thesis, Université Grenoble Alpes (ComUE), 2019. http://www.theses.fr/2019GREAT045.
While Moore’s law reaches its limits, microelectronics actors are looking for new paradigms to ensure future developments of our information society. Inspired by biologic nervous systems, neuromorphic engineering is providing new perspectives which have already enabled breakthroughs in artificial intelligence. To achieve sufficient performances to allow their spread, neural processors have to integrate neuron circuits as small and as low power(ed) as possible so that artificial neural networks they implement reach a critical size. In this work, we show that it is possible to reduce the number of components necessary to design an analogue spiking neuron circuit thanks to the functionalisation of parasitic generation currents in a BIMOS transistor integrated in 28 nm FD-SOI technology and sized with the minimum dimensions allowed by this technology. After a systematic characterization of the FD-SOI BIMOS currents under several biases through quasi-static measurements at room temperature, a compact model of this component, adapted from the CEA-LETI UTSOI one, is proposed. The BIMOS-based leaky, integrate-and-fire spiking neuron (BB-LIF SN) circuit is described. Influence of the different design and bias parameters on its behaviour observed during measurements performed on a demonstrator fabricated in silicon is explained in detail. A simple analytic model of its operating boundaries is proposed. The coherence between measurement and compact simulation results and predictions coming from the simple analytic model attests to the relevance of the proposed analysis. In its most successful achievement, the BB-LIF SN circuit is 15 µm², consumes around 2 pJ/spike, triggers at a rate between 3 and 75 kHz for 600 pA to 25 nA synaptic currents under a 3 V power supply
Causo, Matteo. "Neuro-Inspired Energy-Efficient Computing Platforms." Thesis, Lille 1, 2017. http://www.theses.fr/2017LIL10004/document.
Big Data highlights all the flaws of the conventional computing paradigm. Neuro-Inspired computing and other data-centric paradigms rather address Big Data to as resources to progress. In this dissertation, we adopt Hierarchical Temporal Memory (HTM) principles and theory as neuroscientific references and we elaborate on how Bayesian Machine Learning (BML) leads apparently totally different Neuro-Inspired approaches to unify and meet our main objectives: (i) simplifying and enhancing BML algorithms and (ii) approaching Neuro-Inspired computing with an Ultra-Low-Power prospective. In this way, we aim to bring intelligence close to data sources and to popularize BML over strictly constrained electronics such as portable, wearable and implantable devices. Nevertheless, BML algorithms demand for optimizations. In fact, their naïve HW implementation results neither effective nor feasible because of the required memory, computing power and overall complexity. We propose a less complex on-line, distributed nonparametric algorithm and show better results with respect to the state-of-the-art solutions. In fact, we gain two orders of magnitude in complexity reduction with only algorithm level considerations and manipulations. A further order of magnitude in complexity reduction results through traditional HW optimization techniques. In particular, we conceive a proof-of-concept on a FPGA platform for real-time stream analytics. Finally, we demonstrate we are able to summarize the ultimate findings in Machine Learning into a generally valid algorithm that can be implemented in HW and optimized for strictly constrained applications
Abderrahmane, Nassim. "Impact du codage impulsionnel sur l’efficacité énergétique des architectures neuromorphiques." Thesis, Université Côte d'Azur, 2020. http://www.theses.fr/2020COAZ4082.
Nowadays, Artificial Intelligence (AI) is a widespread concept applied to many fields such as transportation, medicine and autonomous vehicles. The main AI algorithms are artificial neural networks, which can be divided into two families: Spiking Neural Networks (SNNs), which are bio-inspired models resulting from neuroscience, and Analog Neural Networks (ANNs), which result from machine learning. The ANNs are experiencing unprecedented success in research and industrial fields, due to their recent successes in many application contexts such as image classification and object recognition. However, they require considerable computational capacity for their deployment which is not adequate to very constrained systems such as 'embedded systems'. To overcome these limitations, many researchers are interested in brain-inspired computing, which would be the perfect alternative to conventional computers based on the Von Neumann architecture (CPU/GPU). This paradigm meets computing performance but not energy efficiency requirements. Hence, it is necessary to design neuromorphic hardware circuits adaptable to parallel and distributed computing. In this context, we have set criteria in terms of accuracy and hardware implementation cost to differentiate the two neural families (SNNs and ANNs). In the case of simple network topologies, we conducted a study that has shown that the spiking models have significant gains in terms of hardware cost when compared to the analog networks, with almost similar prediction accuracies. Therefore, the objective of this thesis is to design a generic neuromorphic architecture that is based on spiking neural networks. To this end, we have set up a three-level design flow for exploring and implementing neuromorphic architectures.In an energy efficiency context, a thorough exploration of different neural coding paradigms for neural data representation in SNNs has been carried out. Moreover, new derivative versions of rate-based coding have been proposed that aim to get closer to the activity produced by temporal coding, which is characterized by a reduced number of spikes propagating in the network. In this way, the number of spikes can be reduced so that the number of events to be processed in the SNNs gets smaller. The aim in doing this approach is to reduce the hardware architecture's energy consumption. The proposed coding approaches are: First Spike, which is characterized using at most one single spike to present an input data, and Spike Select, which allows to regulate and minimize the overall spiking activity in the SNN.In the RTL design exploration, we quantitatively compared three SNN architectural models having different levels of computing parallelism and multiplexing. Using Spike Select coding results in a distribution regulation of the spiking data, with most of them generated within the first layer and few of them propagate into the deep layers. Such distribution benefits from a so-called 'hybrid architecture' that includes a fully-parallel part for the first layer and multiplexed parts to the other layers. Therefore, combining the Spike Select and the Hybrid Architecture would be an effective solution for embedded AI applications, with an efficient hardware and latency trade-off.Finally, based on the architectural and neural choices resulting from the previous exploration, we have designed a final event-based architecture dedicated to SNNs supporting different neural network types and sizes. The architecture supports the most used layers: convolutional, pooling and fully-connected. Using this architecture, we will be able to compare analog and spiking neural networks on realistic applications and to finally conclude about the use of SNNs for Embedded Artificial Intelligence
Sa'ad, Aisha. "Developing integrated maintenance strategies for renewable energy sources based on analytical methods and artificial intelligence (AI) : comparisons and case study." Electronic Thesis or Diss., Université de Lorraine, 2023. http://www.theses.fr/2023LORR0080.
The development of renewable energy, especially solar and wind energy, over the recent years has gained global attention as an alternative method of generating energy experiencing exceptional growth in its production. In The Global Energy report, global solar energy is expected to have reached a cumulative capacity of 1TW while the wind energy is expected to have multiplied up to 3 to 4 times from mega production in the year 2020. This increase in the solar and wind power implies very significant financial investments. However, with this huge investment potential and significant increase in generation capacity, there is an additional, often overlooked responsibility: managing the power plants to ensure the lowest total life cycle cost (Life Cycle Cost). Like any standard production system, renewable energy (solar and wind energy in our case) generation components are subject to random failure, which interrupts production and supply of demand. Maintenance is identified as a major cause of accidents, lack of technical know-how of an equipment and the absence of a good maintenance routine plan. As part of the efforts to improve the efficiency and performance of renewable energy power plants, we propose models to optimize the power production and maintenance of our selected case studies (Sokoto solar plant and Katsina wind farm). In this regard, we developed new integrated maintenance policies integrated with production of the energy production from solar and wind energy systems. The preventive maintenance strategy adopted in this thesis is perfect maintenance strategy on the selected components for maintenance and an imperfect selective maintenance on the system (solar PV and wind turbine). Battery shortage in case of under-production and maintenance losses are challenges considered in this study. The methodology we developed entails solving the problem of energy production and maintenance optimization by using the theoretical method as well as machine learning method (ANN and SVM) in order to satisfy a random demand of energy during a finite horizon. We also studied the influence of environmental and operational condition of the systems and then validated the models by numerical examples and sensitivity studies proving the robustness of the developed models
Lamy, François. "Étudier le polyusage récréationnel de drogues à travers une simulation multi-agents ontologique." Thesis, Lille 1, 2013. http://www.theses.fr/2013LIL12027/document.
This thesis investigates the career of recreational polydrug users through a pluridisciplinary perspective. This perspective captures the complexity of this phenomenon by integrating data from neurology with a sociological approach. These theoretical elements are integrated into a multi-agent model aiming to test this approach and extend its results. To inform the model, thirty-eight qualitative interviews were conducted with socially-integrated polyusers. After a first phase where drug consumption is oriented toward peers group integration and during which consumption techniques are learnt, the users tend to instrument drugs to facilitate their adaptation to modern social norms and manage social constrains. The polyconsumption appears to be the climax of this psychoactive substances instrumenting and could take four forms permitting the users to make vary their physical and psychological states at will. The last phase of the career is characterized by an increase in control techniques allowing individuals to conciliate their consumptions with the increase of their daily obligations. The status of controller user is defined by opposition to the stereotype of the dependant user, which participates to the labeling of these latter as deviant users. These empirical results have been formalized through visual diagrams before being implemented into the NetLogo platform. The model created, called SimUse, was verified by the means of several scenarios assessing the consistency between the implemented algorithms and collected empirical data
García, Durán Alberto. "Learning representations in multi-relational graphs : algorithms and applications." Thesis, Compiègne, 2016. http://www.theses.fr/2016COMP2271/document.
Internet provides a huge amount of information at hand in such a variety of topics, that now everyone is able to access to any kind of knowledge. Such a big quantity of information could bring a leap forward in many areas if used properly. This way, a crucial challenge of the Artificial Intelligence community has been to gather, organize and make intelligent use of this growing amount of available knowledge. Fortunately, important efforts have been made in gathering and organizing knowledge for some time now, and a lot of structured information can be found in repositories called Knowledge Bases (KBs). A main issue with KBs is that they are far from being complete. This thesis proposes several methods to add new links between the existing entities of the KB based on the learning of representations that optimize some defined energy function. We also propose a novel application to make use of this structured information to generate questions in natural language
Hedayat, Sara. "Conception et fabrication de neurones artificiels pour le traitement bioinspiré de l'information." Thesis, Lille 1, 2018. http://www.theses.fr/2018LIL1I039/document.
Current computing technology has now reached its limits and it becomes thus urgent to propose new paradigms for information processing capable of reducing the energy consumption while improving the computing performances. Moreover, the human brain, is a fascinating and powerful organ with remarkable performances in areas as varied as learning, creativity, fault tolerance. Furthermore, with its total 300 billion cells, is able to perform complex cognitive tasks by consuming only around 20W. In this context, we investigated a new paradigm called neuromorphic or bio-inspired information processing.More precisely, the purpose of this thesis was to design and fabricate an ultra-low power artificial neuron using recent advances in neuroscience and nanotechnology. First, we investigated the functionalities of living neurons, their neuronal membrane and explored different membrane models known as Hodgkin Huxley, Wei and Morris Lecar models. Second, based on the Morris Lecar model, we designed analog spiking artificial neurons with different time constants and these neurons were fabricated using 65nm CMOS technology. Then we characterized these artificial neurons and obtained state of the art performances in terms of area, dissipated power and energy efficiency. Finally we investigated the noise within these artificial neurons, compared it with the biological sources of noise in a living neuron and experimentally demonstrated the stochastic resonance phenomenon. These artificial neurons can be extremely useful for a large variety of applications, ranging from data analysis (image and video processing) to medical aspect (neuronal implants)
Atouati, Samed. "Du texte aux chiffres : La NLP pour la prédiction des actifs financiers." Electronic Thesis or Diss., Institut polytechnique de Paris, 2022. http://www.theses.fr/2022IPPAT026.
The goal of the thesis is to investigate whether natural language data can be useful in better understanding the relationships between the public and the companies, as well as the public and the stock market price changes. In order to do so, we investigate natural language data derived from various sources: Twitter, company filings, and Reddit. We show case promising results for some sources, while others happen to have limited use as far as stock price changes go. Although they remain relevant for understanding public’s reactions to company scandals
Meguenani, Anis. "Safe control of robotic manipulators in dynamic contexts." Thesis, Paris 6, 2017. http://www.theses.fr/2017PA066420/document.
The intended goal of this thesis is to bring new insights for developing robotic systems capable of safely sharing their workspace with human-operators. Within this context, the presented work focuses on the control problem. The following questions are tackled:-for reactive control laws, i.e., control problems where the task to be performed is not known in advance but discovered on-line, how is it possible to guarantee for every time-step the existence of a solution to the control problem? This solution should allow the robot to accomplish at best its prescribed task and at the same time to strictly comply with existing constraints, among which, constraints related to the physical limitations of its actuators and joints.-How to integrate the human-operator in the control loop of the robot so that physical contact can safely be engaged and de-engaged? Regarding the first point, our work arises as the continuity of previous results developed by Sébastien Rubrecht during his PhD thesis. Sébastien Rubrecht introduced the concept of constraints incompatibility for robots reactively controlled at the kinematic-level. The problem of constraints incompatibility appears for example when the formulation of the constraint on an articular position of a robot does not account for the amount of deceleration producible by its actuator. In such case, if the articular position constraint is activated tardively, the system may not have sufficient time to cope with the imposed joint position limit considering its bounded dynamic capabilities
Bou, Tayeh Gaby. "Towards smart firefighting using the internet of things and machine learning." Thesis, Bourgogne Franche-Comté, 2020. http://www.theses.fr/2020UBFCD015.
In this thesis, we present a multilevel scheme consisting of both hardware and software solutions to improve the daily operational life of firefighters. As a core part of this scheme, we design and develop a smart system of wearable IoT devices used for state assessment and localization of firefighters during interventions. To ensure a maximum lifetime for this system, we propose multiple data-driven energy management techniques for resource constraint IoT devices. The first one is an algorithm that reduces the amount of data transmitted between the sensor and the destination (Sink). This latter exploits the temporal correlation of collected sensor measurements to build a simple yet robust model that can forecast future observations. Then, we coupled this approach with a mechanism that can identify lost packets, force synchronization, and reconstruct missing data. Furthermore, knowing that the sensing activity does also require a significant amount of energy, we extended the previous algorithm and added an additional adaptive sampling layer. Finally, we also proposed a decentralized data reduction approach for cluster-based sensor networks. All the previous algorithms have been tested and validated in terms of energy efficiency using custom-built simulators and through implementation on real sensor devices. The results were promising as we were able to demonstrate that our proposals can significantly improve the lifetime of the network. The last part of this thesis focusses on building data-centric decision-making tools to improve the efficiency of interventions. Since sensor data clustering is an important pre-processing phase and a stepstone towards knowledge extraction, we review recent clustering techniques for massive data management in IoT and compared them using real data for a gas leak detection sensor network. Furthermore, with our hands on a large dataset containing information on 200,000 interventions that happened during a period of 6 years in the region of Doubs, France. We study the possibility of using Machine Learning to predict the number of future interventions and help firefighters better manage their mobile resources according to the frequency of events