6558 research outputs found
Sort by
Recurrent Neural Networks model for injury prevention within a professional rugby union club: a proof of concept over one season
Full text linkBackground
In professional rugby, injury prevention and player availability are major challenges. Sports analytics use data from trainings and matches to address these issues. This study leveraged comprehensive daily data from a professional rugby club to predict players' readiness for training. Using this metric helped assess its effectiveness in predicting intrinsic injuries and improving injury prevention strategies.
Methods
Models including logistic regression, decision trees, and Long Short-Term Memory-based neural networks, were evaluated for their predictive accuracy and ability to discern patterns indicative of injury risks or readiness for physical activities.
Findings
The study demonstrated that long-short term memory and convolutional one-dimension models outperform traditional machine learning methods in analyzing players' physical conditions. This approach may support earlier identification of injury risks and inform workload management. Using model evaluation and interpretability techniques, including Local Interpretable Model-Agnostic Explanations (LIME) module, the study provided a framework for sports scientists, coaches, and medical staff to mitigate injury risks and optimize training sessions.
Interpretation
As a preliminary exploration, this study paves the way for further research into the integration of machine learning and neural networks in sports science, promising transformative impacts on injury prevention strategies in rugby
Why move during virtual reality sketching? Experimental study to improve the quality of sketches in virtual reality
Full text linkVirtual Reality (VR) sketching is a valuable tool for conceptual understanding, creativity, and design, but quality issues can hinder its adoption. To address this, we conducted a study involving 15 novices and 15 experts who sketched three chair models in static, mobile, and control conditions. The results showed that mental rotation skills, training, model type, and movement impact sketch quality. The static condition negatively affected performance, particularly volume and proportion. Conversely, the mobile condition didn't improve sketch quality compared to the control group. 3D perception seems tied to movement, highlighting the need to adapt VR sketching software for these challenges. Enhancing the user experience and addressing these quality concerns will be pivotal in the widespread acceptance of VR sketching tools
Single atom convolutional matching pursuit: Theoretical framework and application to Lamb waves based structural health monitoring
Full text linkLamb Waves (LW) based Structural Health Monitoring (SHM) aims to monitor the health state of thin structures. An Initial Wave Packet (IWP) is sent in the structure and interacts with boundaries, discontinuities, and with eventual damages thus generating many wave packets. An issue with LW based SHM is that at least two LW dispersive modes simultaneously exist. Matching Pursuit Method (MPM), which approximates a signal as a sum of delayed and scaled atoms taken from a known dictionary, is limited to nondispersive signals and relies on a priori known dictionary and is thus inappropriate for LW-based SHM. Single Atom Convolutional MPM, which addresses dispersion by decomposing a signal as delayed and dispersed atoms and limits the learning dictionary to only one atom, is alternatively proposed here. Its performances are demonstrated on numerical and experimental signals and it is used for damage monitoring. Beyond LW-based SHM, this method remains very general and applicable to a large class of signal processing problems
Multi-scale optimisation of variable-stiffness composites for thermal cloak
No full textVariable-stiffness composites (VSCs) can be efficiently designed, through multi-scale optimisation, to obtain thermal cloaks that can steer the heat flux to conceal the presence of an obstacle. Specifically, a general class of VSC structures characterised by variable fibres volume fraction, thickness and orthotropy orientation is considered in this work. The theoretical/numerical framework relies on the use of the polar formalism to describe the anisotropic thermal conductivity tensor of the VSC at the macroscopic scale, and on a general numerical homogenisation method to set the link between the design variables and the physical responses defined at different scales. In this context, the goal is to determine the optimal distribution of the fibres volume fraction, the orientation of the main orthotropy axis and the thickness of the VSC structure in order to design an efficient thermal cloak. The design variables fields are represented through non-uniform rational basis spline (NURBS) entities in order to achieve solutions that are compliant with standard computer-aided design software. Moreover, some properties of the NURBS entities, such as the local support property, are conveniently exploited to formally derive the gradient of the physical responses (and hence to speed-up the optimisation process), to automatically satisfy some manufacturability constraints (e.g., the continuity of the fibres-path) and to obtain mesh-independent solutions. The general nature of the proposed approach enables the concurrent optimisation of geometrical and physical variables at multiple scales, thus allowing the identification of optimised solutions that overcome the inherent limitations of the analytical solutions. The effectiveness of this approach is tested on benchmark problems taken from the literature. This entails an investigation into the impact of various factors, including the initial guess, the microscopic configuration of the constitutive phases of the composite material, the boundary conditions, the local thickness, the size and shape of the design region on the optimised solution
Energy velocity of elastic guided waves in immersed plates for complex frequencies and slownesses
Full text linkThe computation of guided modes in fluid-loaded multilayer plates is generally done by a spatial approach, i.e. solutions are sought for a complex slowness. An alternative approach, less frequently employed, involves seeking solutions for complex frequencies. These frequencies correspond to plate resonances. They denote transient phenomena and the guided modes exhibit non-harmonic behavior. Consequently, conventional methods of averaging over time periods become unsuitable for calculating the means of energy quantities. In other words, the calculation of average fields cannot be reduced to a single average over a time period. To tackle this issue, for a predetermined mode, the average fields are obtained through a single averaging process applied to an arbitrary phase term. This averaging process renders independent the means of all energy quantities from the arbitrary origin phase. As usual, an additional integration across the thickness is conducted to derive total energy quantities. Doing this, the total average fields depend on both time and position on the surface plate. A set of four equations is derived from instantaneous and local energy balance equations. From these averages, the energy velocity can be directly calculated. The equations provide further insights into wave dispersion and damping along the energy flow direction, arising from viscoelastic losses and leakages in fluid
Implementation of the sliding-mode controller on overhead cranes via the deadbeat method
No full textThis paper addresses the controller design for the overhead cranes. To this end, a nonlinear model of the overhead cranes is considered, and a continuous-time sliding-mode controller is designed for such a model, ensuring global robust stability. Subsequently, the deadbeat implementation method is employed to obtain a discrete-time equivalent of the designed controller, which is a crucial step to implement any continuous-time controller on digital processors. Comparative analyses based on numerical simulations show that the developed sliding-mode controller shows several advantages over the forward Euler discretization method, which is widely used in the literature
Contribution à la mise en œuvre de l’économie circulaire pour les Matières Plastiques (MP) pour la REP ASL
Full text linkLa Responsabilité Elargie des Producteurs (REP) pour les Articles de Sport et de Loisirs (ASL) a été instaurée en France depuis le 1er janvier 2022. ..
Mechanical properties of decellularized porcine esophagus: Preliminary results
Full text linkEsophageal tissue engineering is a promising approach to create an esophageal substitute after surgical resection of a part of the organ. Regeneration of esophageal tissue may be achieved using some synthetic or biological scaffolds. In the present study, scaffolds are obtained through the decellularization of porcine esophagi. In view of future implantation, it is important to test the mechanical properties of the decellularized matrices and to compare them with the data obtained for native pig esophagi. Results of longitudinal and circumferential traction experiments as well as inflation and burst tests are presented. The results obtained for the compliance of porcine decellularized matrices are novel. It is concluded that the decellularized matrices are suitable for use as esophageal substitutes
COQTEL project dataset : Corrosion Quantification Trough Extended use of Lamb waves
Full text linkCorrosion poses significant safety and cost challenges in the aeronautic industry. Ultrasonic Lamb Waves (LW), emitted and received by a sparse array of piezoelectric elements (PZT), offer an efficient, cost-effective, and versatile solution for corrosion monitoring. This dataset corresponds to two experiments involving a LW solution based on a sparse PZT array and able to monitor corrosion pit growth on a 316L stainless steel plate during controlled corrosion. The corrosion pit size is electrochemically controlled by the imposed electrical potential and the injection of a corrosive NaCl solution through a capillary at the desired pit location. Simultaneously, the corrosion pit growth is monitored in-situ every 10 seconds using a sparse array of 4 PZTs bonded to the back of the steel plate. Two independent experiments were conducted to assess the repeatability of this approach. The collected dataset collected can facilitate the development of Structural Health Monitoring (SHM) algorithms and methodologies, provide data for waves/damage interaction modeling, and help bridging the gap between research and industry in this domain
In-situ monitoring of µm-sized electrochemically generated corrosion pits using Lamb waves managed by a sparse array of piezoelectric transducers
Full text linkCOQTEL project dataset : Corrosion quantification trough extended use of Lamb waves
Data in Brief, Volume 59, April 2025, Pages 111393
C. Nicard, M. Rébillat, O. Devos, M. El May, F. Letellier, S. Dubent, M. Thomachot, M. Fournier, P. Masse, N. MechbalCorrosion is a major threat in the aeronautic industry, both in terms of safety and cost. Efficient, versatile, and cost affordable solutions for corrosion monitoring are thus needed. Ultrasonic Lamb Waves (LW) appear to be very efficient for corrosion monitoring and can be made cost effective and versatile if emitted and received by a sparse array of piezoelectric elements (PZT). A LW solution relying on a sparse PZT array and allowing to monitor µm-sized corrosion pit growth on stainless 316L grade steel plate is here evaluated. Experimentally, the corrosion pit size is electrochemically controlled by both the imposed electrical potential and the injection of a corrosive NaCl solution through a capillary located at the desired pit location. In parallel, the corrosion pit growth is monitored in-situ every 10 s by sending and measuring LW using a sparse array of 4 PZTs bonded to the back of the steel plate enduring corrosion. As a ground truth information, the corrosion pit volume is estimated as the dissolved volume balancing the electronic charges exchanged during corrosion. The corrosion pit radius is additionally checked post-experiment precisely with an optical measurement. Measured LW signals are then post-processed in order to compute a collection of synthetic damage indexes (DIs). After dimension reduction steps, obtained DI values correlates extremely well with the corrosion pit radius. Using a linear model relating those DI values to corrosion pit radius, it is demonstrated that corrosion pit from 30 µm to 150 µm can be reliably detected, located, and their upcoming size extrapolated. Two independent experiments were achieved in order to ensure the repeatability of the proposed approach. LW managed by a sparse PZT array thus appears to be reliable and efficient to monitor growth of µm-sized corrosion pits on 316L steel plates. If embedded in aeronautical structure, such an approach could be a versatile and cost-effective alternative to actual non-destructive maintenance procedures that are time and manpower consuming