Université de Technologie Tarbes Occitanie Pyrénées

Predicting the remaining useful life (RUL) of fast-charging lithium-ion batteries using early-stage lifecycle data is remains challenging due to limited run-to-failure data and lack of knowledge on battery degradation mechanisms. To address this issue, a generic Physics-Informed Machine Learning (PIML) framework is developed. The PIML framework consists of two parallel branches: a physics-informed (PI) branch and a data-driven branch. The PI branch is a neural network stacked by the linear projection layers with embedded physics knowledge, while the data-driven branch is a task-specific machine-learning model. In addition, a three-step training strategy is introduced, including (1) Training the data-driven branch, (2) Training the PI branch for aligning physical consistency without updating the hyperparameters in the data-driven branch, and (3) Fine-tuning both branches simultaneously to achieve optimal performance. To validate this framework, a physics-based model that represents the growth of solid electrolyte interphase (SEI) and a dilated convolutional neural network are implemented in the PI and data-driven branches, respectively. The solid electrolyte interphase-informed dilated convolutional neural network (SEI-DCN) model is demonstrated on the Stanford–MIT–Toyota-battery dataset. Using only four lifecycle data, the SEI-DCN model achieves very high prediction accuracy compared to standard dilated CNNs and other state-of-the-art models under various testing conditions and lifetime ranges. Moreover, the framework is generalizable to different physics-based battery degradation models. • Novel dual-branch parallel PIML framework to merge varied knowledge on battery degradation. • The new learning strategy ensures that PIML is lower bounded by the performance of data-driven models with imprecise knowledge. • Building SEI-informed DCNN validated on the fast-charging lithium-ion batteries and outperformed the SOTA model. • SEI-informed DCNN preserves SOTA prediction accuracy on completely new data with different operation conditions and life spans of the training set. • Investigation of the adaptability of the proposed framework for knowledge replacement across various SEI models.

Despite the well-established nature of non-destructive testing (NDT) technologies, autonomous monitoring systems are still in high demand. The solution lies in harnessing the potential of intelligent structures, particularly in industries like aeronautics. Substantial downtime occurs due to routine maintenance, leading to lost revenue when aircraft are grounded for inspection and repairs. This article explores an innovative approach using intelligent materials to enhance condition-based maintenance, ultimately cutting life-cycle costs. The study emphasizes a paradigm shift toward structural health monitoring (SHM), utilizing embedded sensors for real-time monitoring. Active thin film piezoelectric materials are proposed for their integration into composite structures. The work evaluates passive sensing through acoustic emission (AE) signals and active sensing using Lamb wave propagation, presenting amplitude-based and frequency domain approaches for damage detection. A comprehensive signal processing approach is presented, and the damage index and damage size correlation function are introduced to enable continuous monitoring due to their sensitivity to changes in material properties and defect severity. Additionally, finite element modeling and experimental validation are proposed to enhance their understanding and applicability. This research contributes to developing more efficient and cost-effective aircraft maintenance approaches through SHM, addressing the competitive demands of the aeronautic industry.

Compression-tension asymmetry (CTA) in yield strength is one of the issues that hinders widespread application of wrought Magnesium (Mg) alloys. Strong basal texture formed in thermomechanical processing and the difference in activation of deformation twinning in tension and compression are responsible for CTA. To reduce CTA, the readily implementable Double Equal Channel Angular Pressing (D-ECAP) is used in this study for altering the microstructure and texture of Mg samples. Mg bars containing 1, 5, and 10 wt% Gd were subjected to D-ECAP at 400 °C. Due to plastic strain, the average grain sizes decreased from several hundred microns to about 13.4 μm, 5.6 μm, and 5.1 μm, respectively. All processed samples showed high strength and characteristics shear textures. The crystallographic textures displayed the so-called C1-C2 and B fibers. C1-C2 were the major fibers in the 1 % and 5 % Gd samples, while the B and C1 fibers appeared in the 10 % Gd sample. C1-C2 requires high activity of pyramidal <c + a > slip, and B belongs to basal slip. The CTA was measured by the ratio of the compression/tensile yield stress and significant increase in CTA was obtained on the D-ECAP processed samples with respect to the base alloys. The CTA reached 0.85 in the Mg 10Gd alloy without sacrificing the yield strength. Polycrystal plasticity simulations were done using the experimental textures and good reproduction of the CTA values were achieved. The simulation results also revealed the relative activity of the slip and twinning systems for understanding the mechanisms that control the CTA. The results of this study revealed that D-ECAP is an efficient processing technology that can be widely used in the preparation of high-performance Mg Gd alloys.

Permeability measurement of engineering textiles is a key step in preparing composite manufacturing processes. A radial flow experimental setup was used in this work to measure the unsaturated and saturated in-plane permeabilities of five different types of E-glass textiles and their ratios. In parallel, delayed tow saturation during the oil injection stage was visually observed to identify fabrics that exhibited a significant dual-scale effect. A numerical approach to determine the saturated permeability of a given fabric geometry at the mesoscale was tested and validated against analytical models found in the literature. It was then applied to a realistic geometry acquired from an E-glass plain weave textile using an X-ray microtomography scanner (μCT). Two numerical methods were adopted: the single-scale method, where the tows are considered impermeable, and the dual-scale method, where the permeability of the tows is taken into account. The numerical results from both methods were then compared with the experimental values and showed good agreement, especially with the second method.

Fatigue crack is a critical failure mode in machinery elements, leading to numerous studies focused on understanding crack initiation and propagation mechanisms. Two main methods have been proposed in the literature for crack propagation prediction: data-driven and model-based methods. Data-driven methods are widely used for crack prognostic but demand extensive, high-quality datasets. In contrast, fatigue models can accurately predict component degradation and remaining life under cyclic loading if their parameters are well-aligned with the data. This work proposes a physics-guided framework to improve crack propagation prediction through the Paris law, even with limited datasets, using the Broyden-Fletcher-Goldfarb-Shanno (BFGS) optimization method for model parameter estimation, along with data pre-processing and uncertainty quantification. The optimization was performed using weighted data, while data pre-processing incorporated outlier removal to enhance model reliability. When applying the weighting scheme, an improvement of about 15% in prediction accuracy was observed. Further improvements, along with the implementation of a two-stage outlier detection process and Monte Carlo simulations, resulted in a 20% improvement in accuracy. This study highlights that the integration of data pre-processing, parameter estimation with BFGS, and uncertainty quantification effectively tackles the challenge of limited observation data in crack prognostic, distinguishing it from prior approaches.

ABSTRACT The rapid expansion of hydrogen‐powered technologies has developed the need for lightweight, high‐performance storage solutions. Composite overwrapped pressure vessels (COPVs), particularly Type IV designs featuring polymer liners and carbon‐fiber reinforcements, represent the most widely used means of on‐board hydrogen storage due to their exceptional strength‐to‐weight ratios and fatigue resistance. However, conventional thermosetting epoxies, while offering excellent processability, pose significant recyclability challenges amid escalating sustainability and ecology requirements. This review critically examines the state of the art in composite materials for hydrogen tanks, with a special focus on Elium liquid thermoplastic resin as an eco‐friendly alternative. The principal filament winding techniques, namely wet winding, dry winding, towpreg winding, and thermoplastic prepreg winding with in situ consolidation, are systematically reviewed and comparatively analyzed in terms of their processing characteristics, compatibility with resin systems, and resulting composite quality. Recent advancements in the optimization of filament winding process parameters, such as fiber tension, winding angle, and geometry factors, are critically examined for their direct influence on key performance indicators, including burst pressure (the maximum internal pressure a tank can withstand before catastrophic failure) and structural weight. By synthesizing recent advances in materials, processing parameters, and performance metrics, this review outlines a roadmap toward fully recyclable, high‐integrity hydrogen storage solutions.

Abstract. The quality of parts produced by PBF-LB/M processes, while promising, is often compromised by defects such as gas pores and lack of fusion. In response to the global push towards closed-loop systems for real-time defect detection and correction, in-situ correction during the PBF process is critical. Real-time monitoring enables the immediate detection and mitigation of defects, preventing the propagation of defective layers and preserving the integrity of the final component. The present study proposes a novel, cost-effective approach for real-time monitoring of the PBF process. Eight PBF samples, built in AlSi10Mg material and shaped as cylinders and cubes, contained intentionally created non-lased zones to induce lasing defects in the subsequent layers. In-situ monitoring is provided using an ultrasonic acoustic sensor installed inside the building chamber. The data collection is applied on all samples throughout the building process. The collected data is then processed via the application of Artificial Neural Networks (ANN) model, where temporal and frequency features of the ultrasonic signals are extracted and analyzed, in order to detect and identify the defects. The obtained results show that the ANN achieves a success rate of 92% in both accuracy and F1-score, highlighting its effectiveness in detecting the intentionally introduced defects.

In the context of developing PINN methods for real-time digital twins in manufacturing processes, we propose a new approach that combines two complementary weighting strategies to significantly improve their convergence. The first method, called SD-PINN, balances the loss terms associated with the governing equations, boundary conditions, and initial conditions, ensuring that their contributions are dimensionally consistent and therefore comparable in magnitude. The second method, called SDFEET-PINN, rescales the terms of the governing equations during the early stages of training. This facilitates learning by temporarily modifying the equations to make terms comparable in amplitude, and then progressively restoring the original formulation, thereby preserving the influence of lower-magnitude terms that are often neglected in standard PINN approaches. We apply these methods to transient thermal problems, which are critical for predicting defects in Powder Bed Fusion (PBF). A range of 2D configurations with complex boundary conditions is used to test robustness, and a practical case study is carried out on heat transfer in a complex 3D geometry previously investigated both numerically and experimentally in PBF. Results show that the combined SD-PINN and SDFEET-PINN approach achieves higher predictive accuracy and stability compared to classical PINNs. Furthermore, we introduce an Adaptive Learning Rate strategy that reduces the step size after initial stabilization, further enhancing predictive performance and enabling efficient convergence across all test cases.

The adaptability of robotic systems is expanding the horizons of manufacturing flexibility. However, fully leveraging the potential of these systems poses considerable challenges. A key requirement is the ability to understand and model their diverse capabilities through a standardized and semantically well-defined framework. In this paper, we introduce the Robotic Capability Ontology (RCO), developed through a systematic investigation of various types of robotic capabilities, including those related to function, quality, and process performance. We define two types of capabilities: Advertised capabilities, as specified by manufacturers, and Operational capabilities, which reflect real-world performance. The RCO framework provides an ontology-based approach to representing these capabilities in a structured and interpretable manner. Within the manufacturing context, RCO serves as a reference ontology that bridges manufacturer specifications and empirical performance data to support more accurate, explainable, and interoperable representations of robotic capabilities.

This study deals with electromagnetic characterization of power electronic modules wire bonding failure. The several steps of the characterization method in order to reveal high frequency signature, are detailed. The application of the method is illustrated on isolated prototype. Simulation and experimental results highlight the same signature of the failure.

International audience

International audience

Nowadays, the integration of power electronics into power systems has increased significantly. As a result, power grids were subject to numerous transformations. Hence, operations, power factor corrections, and voltage control of the grid-connected converters have become challenging tasks subject to enhancing the stability, reliability, and performance of the power grid. For this, we propose in this paper an advanced state-feedback control using the linear quadratic regulator (LQR) optimized by the metaheuristic method for a 2-level 3-phase voltage source converter (VSC) connected to the utility grid through an output L-type or LCL-type filter. The proposed methodology avoids the empirical trial-and-error technique for adjusting the weighting values of matrices Q and R respecting multi-objectives: (i) minimize control loop errors, (ii) minimize the overshoot, and (iii) respect suitable time constant. Thus, the optimal weighting matrices are provided using the multi-objectives Tunicate Swarm Algorithm-TSA. The effectiveness of the proposed methodology is tested on Electromagnetic Transients (EMT) VSC grid-connected systems and compared to standard vector control and advanced control (H<inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">∞</inf>). The proposed methodology permits important performance regulation of reactive power for the feasible VSC operation range and under perturbation conditions (short circuits,…).

National audience

ABSTRACT Transportation systems are increasingly evolving into highly integrated cyber‐physical environments that demand advanced coordination between computational intelligence and physical infrastructure. In response to growing requirements for sustainability, resilience and adaptability, this paper proposes a cyber‐physical transportation system (CPTS) framework that prioritises real‐time data‐driven decision‐making. The framework captures the dynamic interactions between cyber components (e.g., sensing, computation and communication) and physical subsystems (e.g., infrastructure, vehicles and users), thereby enabling continuous monitoring, anomaly detection and adaptive control. At its core lies a meta‐architectural design developed through structured knowledge elicitation and operationalised via a design structure matrix (DSM) clustering algorithm to manage system complexity and behavioural evolution. This approach strengthens the system's capacity to detect failures and uncover latent mobility demands, ultimately supporting strategic transportation planning. A real‐world case study validates the framework's effectiveness in guiding policymakers and stakeholders towards the creation of intelligent, sustainable and responsive transportation infrastructures. The findings underscore the transformative potential of CPTS in advancing transportation decision‐making through integrated cyber‐physical reasoning and adaptive system design.

This article introduces a method for enhancing territorial risk management through dysfunctional analysis supported by the SADT tool. It discusses the importance of functional analysis in identifying risks within a territory, emphasizing the need for a comprehensive approach to understanding the various functions and activities within a territory. By applying these principles to emergency response plans the article highlights the use of incidence matrices in risk assessment and process modeling to improve preparedness and response to natural or anthropogenic disasters.

Current specimen designs for biaxial tension-tension fatigue tests of composites are not optimal and often generate unexpected failure in fatigue tests, due to the lack of a design standard and the complexity of composite materials and multiaxial loads. And designing composite structures with uniaxial testing is not sufficient due to the multiaxiality of the stress tensor. This study aims to present an optimized specimen for multidirectional carbon fiber reinforced composites in biaxial tension-tension fatigue tests, in which thermodynamic phenomena occurring inside the specimen will be monitored by infrared thermography. A feasible specimen could be designed by numerical simulation using the finite element method. Firstly, a design criterion is proposed to ensure a failure in the gauge region, a practical manufacturing approach, and optimal conditions for temperature measurement. An initial simulation model is established to find out the proper shape, followed by a more comprehensive simulation used for determining the dimension of the specimen. Consequently, the optimized specimen is designated as a cruciform shape with a reduced gauge region. Last, two stacking sequences of specimens, referred to as cross-ply [(0/90) 6 ] s and quasi-isotropic [(0/45/90/-45) 3 ] s , are tested in a simulation model that takes into account biaxial static and fatigue loadings, and then a thermal and fatigue simulations will be performed to validate the geometry. The favorable simulation results indicate that the optimized specimen and design approach is well-suited for multidirectional composites in biaxial fatigue tests with temperature monitoring.

International audience

This paper presents a novel methodology for bearing fault diagnosis under non-stationary operating conditions using angular resampling combined with sparse representation classification. The proposed approach addresses variable speed challenges by transforming time-domain vibration signals into the angular domain through encoder-based resampling, enabling extraction of speed-invariant envelope order spectrum features. For classification, a structured dictionary is constructed from training samples for each bearing condition (healthy, inner race, outer race, ball, and combined faults). Sparse coding is performed using the fast iterative shrinkage-thresholding algorithm (FISTA) within an ℓ 1 -norm regularized framework, with final class assignment determined by minimizing reconstruction error across class-specific dictionaries. The proposed framework achieved 99.86% cross-validated accuracy on the Ottawa dataset and demonstrated superiority over state-of-the-art methods (classical machine learning classifiers and some deep learning architecture), all evaluated on identical speed-invariant features. Comprehensive cross-domain validation confirmed robust generalization: cross-speed-profile experiments achieved 99.6% mean accuracy when testing on unseen speed dynamics, while cross-load validation achieved 98.8% mean accuracy across different bearing types and loading conditions. All cross-domain scenarios exceeded 97.8% accuracy. This integrated framework provides a transparent, physically interpretable solution for bearing fault diagnosis, offering enhanced robustness to speed variability and load variations, with practical applicability for industrial condition monitoring under realistic operating conditions.

In this paper, we analyze the interoperability landscape for materials and manufacturing in a broad sense and with a particular focus in the context of data. To set the stage and give an overview of the various facets of this topic, we collect and compare existing definitions and classifications of interoperability (its types, layers, levels) and summarize recommendations from various entities and communities. After this, we carry out an analysis on a set of interoperability scenarios, propose a broad structure of requirements for interoperability, and list some key components that can be used to meet these requirements, with a particular emphasis on the role of semantic technologies and knowledge representation. Finally, we highlight future challenges and suggest directions for best practices. Throughout the paper, we emphasize common points and differences in the landscape. Through this process, relevant dimensions are identified, and various tables are provided with useful syntheses and structured categorizations that can serve as a base for future theoretical work, as well as immediate practical guidance (e.g., for requirements gathering, literature navigation).