HAL Portal IMT Nord Europe
Not a member yet
6687 research outputs found
Sort by
Stochastic model predictive control of an irrigation canal with integrated performance-driven path planning of a measurement robot
No full textInternational audienceABSTRACT This work proposes a stochastic model predictive control for an irrigation canal with uncertainties where a moving robot takes measurements across the canal considering criteria such as the robot’s velocity, energy consumption, and distances between the measuring spots. Tightened constraints are applied over the prediction horizon to the optimization so that the controller selects the optimal route for the robot from a control viewpoint. The simulations compare three different approaches, demonstrating that the proposed technique achieves superior results by reducing constraints violations and operational costs and ensuring more precise and reliable water level management across the canal compared to other methods
Présentation de l'infrastructure de recherche ACTRIS-FR et de ses services
No full textInternational audienceACTRIS-FR constitue la contribution nationale française à ACTRIS (Aerosol, Clouds and Trace Gases Research Infrastructure), une infrastructure de recherche européenne distribuée dédiée à l'observation et à l'étude des aérosols, des nuages et des gaz réactifs, ainsi qu'à leurs interactions. ACTRIS joue un rôle essentiel dans le soutien des recherches sur le climat et la qualité de l'air, en fournissant des données et des services accessibles aux chercheurs et aux acteurs du domaine. Ce poster présente les services proposés par ACTRIS-FR, notamment l'accès aux données issues de l'infrastructure, ainsi que les projets phares dans lesquels ACTRIS-FR est impliquée
Exploring the serious game ‘What's going on in the air?’: An innovative tool bridging atmospheric research and active science education
No full textInternational audienc
Methodologies to Design Optimum 3D Printable Mortar Mix: A Review
No full textInternational audienceNowadays, 3D printing has revolutionized the construction and building industry, enabling researchers to push the boundaries of creating structural components with this innovative technique. A key factor for the success of this approach lies in selecting the optimal mix design, which must possess suitable properties for printing while ensuring strong performance once hardened. However, achieving this optimal mix is complex due to limited knowledge regarding the necessary fresh-state properties, the characteristics and proportions of the constituents, the influence of printing parameters on these properties, and the various challenges encountered during and post printing. This paper aims to address these aspects by offering a comprehensive review of the steps researchers have taken to develop an optimized 3D printable mix
Towards the validation of an in vitro method to predict the bioavailability of soil organic contaminants
No full textInternational audienceFor several decades, the presence of organic contaminants such as polycyclic aromatic hydrocarbons (PAHs), polychlorinated biphenyls (PCBs) or pesticides such as chlordecone (CLD) in ecosystems has raised significant concerns regarding environment and human health. Humans may be exposed to these substances through unintentional ingestion of soil (hand-to-mouth contact particularly common in children [1]), potentially leading to health hazards. Considering the total contaminant concentration in soil may lead to an overestimation of the risk, as only the bioaccessible fraction released into the digestive tract can be absorbed. Predicting the bioavailable fraction by estimating oral bioaccessibility is relevant for accurate risk assessment [2]. Bioaccessibility can be estimated by in vitro methods simulating human digestion, already available for trace metal elements (for ex. Unified Bioaccessibility Method). The PBET (Physiology Based Extraction Test) method simulates a stomachal digestion following by an intestinal digestion. Recently, an adaptation for hydrophobic organic contaminants has been suggested by adding Tenax (absorbent sink) in the intestinal phase, improving the extraction of organics liberated from soils (for ex. DDT pesticide [3], PCBs [4], PAHs [5]). The Ti-PBET (Tenax-improved PBET) method shows promising results, but its use may be limited by the high cost of Tenax. The general objective of our work was therefore to improve the Ti-PBET test in terms of both practicability and cost. We compared different adsorption supports for extracting PAHs and PCBs, and found XAD-2 performed comparably to Tenax while being more cost-effective; XAD-2 was therefore selected for subsequent test improvements. To simplify resin recovery after intestinal digestion, we tested two ways of incorporating XAD-2 into the in vitro assay: (1) enclosed in a tea bag; (2) freely dispersed within the flask. No significant differences were observed between the two types of control samples (p > 0.05), although certain compounds such as PCB 28 and fluorene showed higher chromatogram signals with the tea bag method. Although more convenient, this option was abandoned as it increased the risk of control contamination and compromised the accuracy of bioaccessibility estimates. We finally tested the Xi-PBET (XAD-2–improved PBET) method by spiking PCBs at 1, 5, 10, 50, and 200 ng/mL (n = 3) to assess the recovery rates of six congeners (PCB 28, 52, 101, 138, 153, 180) and GC–MS/MS quantification accuracy across this concentration range. Recovery rates ranged from 87–132% (CV ≤ 12%) between 5 and 200 ng/mL, while the 1 ng/mL concentration showed higher overestimation and variability (120–188%; CV ≤ 69%) likely due to low spiking volumes and background noise. These results confirm that 500 mg of XAD-2 effectively captures the entire free PCB fraction between 5 and 200 ng/mL without saturation of the resin, and that GC–MS/MS is reliable for this concentration range. This work has demonstrated that XAD-2 resin is a good alternative to Tenax for reducing the cost of bioaccessibility testing without compromising its effectiveness, making such testing more accessible to the contaminated sites and soils sector. However, in vitro bioaccessibility methods still require improvement to provide reliable estimates. Therefore, our future work will compare Xi-PBET against other in vitro assays using a wide range of soils contaminated with different pollutant families, and validate results against bioavailability measurements obtained from in vivo experiments on piglets (the best model to simulate children’s physiology). The in vitro methodologies are expected to provide a valuable tool for site and soil managers in health risk assessmen
Inverted BER Trends for Energy-Detected GRSM-MQAM Massive MIMO Downlink
No full textNational audienceThis work investigates Generalized Receiver Spatial Modulation (GRSM) in a massive MIMO downlink scenario over millimetre wave channels. While GRSM enhances spectral efficiency (SE) and reduces power consumption, indexing additional bits using the spatial dimension increases the vulnerability to detection errors. These errors primarily stem from thresholddependent spatial detection. We propose a novel predefined threshold computation method minimizing spatial detection errors, rigorously validated through the Maximum A Posteriori (MAP) criterion. Furthermore, we derive an analytical Average Bit Error Probability (ABEP) expression tailored for energy detection, exploiting inherent constellation energy distributions. The analytical derivation was validated via link-level simulations under two scenarios: (i) perfect spatial detection, and (ii) practical spatial detection. The results show spatial errors dominating overall performance, shifted from theoretical values by practical spatial detection using multiple thresholds for different energy levels in 16QAM. Crucially, an inverted error trend revealed between GRSM-4QAM and GRSM-16QAM, highlighting a tradeoff between error resilience, complexity, and energy efficiency
Measuring Anxiety Levels with Head Motion Patterns in Severe Depression Population
No full textInternational audienceDepression and anxiety are prevalent mental health disorders that frequently cooccur, with anxiety significantly influencing both the manifestation and treatment of depression. An accurate assessment of anxiety levels in individuals with depression is crucial to develop effective and personalized treatment plans. This study proposes a new noninvasive method for quantifying anxiety severity by analyzing head movements -specifically speed, acceleration, and angular displacementduring video-recorded interviews with patients suffering from severe depression. Using data from a new CALYPSO Depression Dataset, we extracted head motion characteristics and applied regression analysis to predict clinically evaluated anxiety levels. Our results demonstrate a high level of precision, achieving a mean absolute error (MAE) of 0.35 in predicting the severity of psychological anxiety based on head movement patterns. This indicates that our approach can enhance the understanding of anxiety's role in depression and assist psychiatrists in refining treatment strategies for individuals.</div
Urban Air Quality Management at Low Cost Using Micro Air Sensors: A Case Study from Accra, Ghana
No full textInternational audienceUrban air quality management is dependent on the availability of local air pollution data. In many major urban centers of Africa, there is limited to nonexistent information on air quality. This is gradually changing in part due to the increasing use of micro air sensors, which have the potential to enable the generation of ground-based air quality data at fine scales for understanding local emission trends. Regional literature on the application of high-resolution data for emission source identification in this region is limited. In this study a micro air sensor was colocated at the Physics Department, University of Ghana, with a reference grade instrument to evaluate its performance for estimating PM2.5 pollution accurately at fine scales and the value of these data in identification of local sources and their behavior over time. For this study, 15 weeks of data at hourly resolution with approximately 2500 data pairs were generated and analyzed (June 1, 2023, to September 15, 2023). For this time period a coefficient of determination (r2) of 0.83 was generated with a mean absolute error (MAE) of 5.44 μg m–3 between the pre local calibration micro air sensor (i.e., out of the box) and the reference-grade instrument. Following currently accepted best practice methods (see, e.g., PAS4023) a domain specific (i.e., local) calibration factor was generated using a multilinear regression model, and when this factor is applied to the micro air sensor data, a reduction, i.e. improvement, in MAE to 1.43 μg m–3 was found. Daily variation was calculated, a receptor model was applied, and time series plots as a function of wind direction were generated, including PM2.5/PM10 ratio scatter and count plots, to explore the utility of this observational approach for local source identification. The 3 data sets were compared (out of the box, domain calibrated, and reference-grade) and it was found that although there were variations in the data reported, source areas highlighted based on these data were similar, with input from local sources such as traffic emissions and biomass burning. As the temporal resolution of observational data associated with these micro air sensors is higher than for reference grade instruments (primarily due to costs and logistics limitations), they have the potential to provide insight into the complex, often hyperlocalized sources associated with urban areas, such as those found in major African cities
Explainable Pattern Learning in Exploring Robust Characteristics in Metaheuristic Design
No full textInternational audienceThe Vehicle Routing Problem (VRP) is a complex optimization problem due to its NP-Hard nature, and it is mostly solved using metaheuristic algorithms. Recent developments in machine learning have demonstrated the potential to improve these approaches by substituting human-crafted designs with data-driven methods. Building on this advance, we examine the role of different characteristics or features in predicting the quality of VRP solutions, identifying several features that consistently serve as strong predictors and could be leveraged in the design of metaheuristic algorithms. We suggest that while feature importance can vary, specific characteristics or features remain reliable predictors across different scenarios