Archive ouverte HAL-LAAS
Not a member yet
12189 research outputs found
Sort by
Framework for effective PV system instrumentation focused on fault diagnosis
No full textInternational audienceA comprehensive framework for instrumentation in photovoltaic (PV) systems is proposed to enhance fault detection accuracy and diagnostic capability across varied PV applications. The framework is structured around four key components: (i) system design considerations, which include PV topology, scale, and sensor placement strategies to maximize detection sensitivity; (ii) data acquisition, detailing sensor selection, sampling rate optimization, and communication protocols adaptable to different configurations; (iii) data management and preprocessing, encompassing storage strategies, data quality control, and normalization pipelines; and (iv) a review of existing monitoring platforms, identifying their limitations for fault-specific diagnostics. Unlike existing standards such as International Electrotechnical Commission (IEC) 61724, which focus on performance monitoring, this framework is explicitly tailored to address diagnostic challenges, offering a fault-oriented perspective on instrumentation design. It provides structured guidelines for aligning spatial resolution, sensor types, and data granularity with the specific needs of fault localization and characterization. The framework also promotes scalable and cost-effective solutions by balancing the trade-offs between instrumentation complexity and diagnostic accuracy. By emphasizing dynamic sampling strategies and preprocessing workflows, the framework supports the development of more responsive and reliable monitoring architectures. This contribution aims to guide practitioners and researchers in improving the resilience, maintainability, and performance of PV systems through more intelligent and diagnosis-ready instrumentation infrastructures
Artificial intelligence in photovoltaic fault diagnosis: A Natural Language-Based Topic-tSNE Fusion analysis
No full textInternational audienceTimely fault detection in photovoltaic systems is critical for ensuring energy efficiency, reliability, and cost-effectiveness. However, the nonlinear and weather-dependent behavior of photovoltaic systems poses challenges for accurate diagnosis. This study presents a large-scale review of 983 scientific publications on artificial intelligence-based photovoltaic fault detection, using a novel methodology called Topic-tSNE Fusion. This approach integrates topic modeling, dimensionality reduction, and expert analysis to extract and visualize dominant research themes. Four key machine learning paradigms are identified: supervised, unsupervised, semi-supervised, and reinforcement learning. Among them, supervised methods, particularly neural networks and support vector machines, are the most frequently applied, showing accuracies above 95% in controlled conditions. The analysis also reveals growing use of semi-supervised and hybrid approaches to overcome data scarcity. Commonly monitored variables include irradiance, voltage, and current, while the most studied faults are shading, open-circuit, and degradation. Several open-access datasets supporting fault diagnosis research are catalogued. Overall, the proposed method enables a more objective and scalable review process and uncovers emerging trends, such as the shift toward lightweight artificial intelligence for edge deployment and frugal diagnostic architectures. The methodology is scalable and adaptable to other domains facing similar challenges in knowledge synthesis and system monitoring
Optimizing PV maintenance: Methods, cleaning frequency, and a selection protocol
No full textInternational audienceDust accumulation significantly reduces the efficiency of Photovoltaic (PV) systems, with energy losses reaching up to 50% in arid and semi-arid regions. This study presents a comprehensive review of PV cleaning methods, analyzing both passive (natural mitigation, coated surfaces, architectural solutions) and active methods (manual cleaning, water-based systems, electromechanical techniques, robotic cleaning, and piezoelectric approaches). A systematic evaluation of their operational principles, effectiveness, and economic implications is conducted, considering environmental constraints and site-specific conditions. A key contribution of this study is the assessment of optimal cleaning frequency, identifying how climatic and geographical factors influence maintenance schedules. Additionally, a novel Strategic PV Cleaning Optimization Method (SPV-COM) is introduced, offering a structured, multi-criteria decision-making framework to compare and rank cleaning methods based on technical performance, economic feasibility, and long-term sustainability. This methodology integrates operational costs and maintenance requirements to ensure an adaptive selection process that aligns with real-world PV installations. The proposed framework is scalable and applicable to diverse PV cleaning technologies, supporting decision-making in both small-scale and large-scale installations. The findings highlight, for example, the economic trade-offs between water-intensive methods and emerging autonomous solutions, as well as the need for region-specific strategies. By addressing critical gaps in prior studies, this work provides a structured approach to optimizing PV maintenance, contributing to improved efficiency, cost reduction, and sustainable energy production
Evaluating Embeddable Language Models in Verbalizing Rule-based Inferences through Justifications
No full textInternational audienceWhile Language Models have shown promising performance, they still struggle with limitations regarding reasoning and are very token-sensitive. In contrast, knowledgebased systems, such as ontologies, allow for provable logically valid reasoning and provide explicit justifications regarding newly inferred knowledge. However, those justifications can be hard to understand for non-expert users given their formal syntax and their length. We investigated if language models could be considered as reliable tools for verbalizing such explanations, thus increasing explainability over reasoning output. This paper presents a reference evaluation of a set of embeddable language models on a task of translation from ontology formatted inferences and justifications into natural language sentences. We show that the order of justifications significantly decreases performance, whereas adding the inference rule as additional context significantly improves performance, leading to more reliable results
Latent Conditioned Loco-Manipulation Using Motion Priors
No full textInternational audienceAlthough humanoid and quadruped robots provide a wide range of capabilities, current control methods, such as Deep Reinforcement Learning, focus mainly on single skills. This approach is inefficient for solving more complicated tasks where high-level goals, physical robot limitations and desired motion style might all need to be taken into account. A more effective approach is to first train a multipurpose motion policy that acquires low-level skills through imitation, while providing latent space control over skill execution. Then, this policy can be used to efficiently solve downstream tasks. This method has already been successful for controlling characters in computer graphics. In this work, we apply the approach to humanoid and quadrupedal loco-manipulation by imitating either simple synthetic motions or kinematically retargeted dog motions. We extend the original formulation to handle constraints, ensuring deployment safety, and use a diffusion discriminator for better imitation quality. We verify our methods by performing loco-manipulation in simulation for the H1 humanoid and Solo12 quadruped, as well as deploying policies on Solo12 hardware. Videos and code are available at https://gepetto.github.io/LaCoLoco
Time for the Automotive Industry to Industrialize Safe CI/CD
No full textInternational audienceThere is a huge commercial potential for the automotive industry to move from large platform projects of several years eventually ending up in a start-of-production (SOP), to frequent updates of a continuously evolving product.To enable such continuous deployment (CD), assurance case generation must be automated and integrated into the continuous integration (CI) pipeline, ensuring each change maintains compliance without excessive manual rework. However, continuous deployment in safety-related systems demands more than just tooling -it requires a product architecture that supports modular, contract-based assurance and organizational readiness for continuous delivery.The path towards true success begins with demonstrating technical feasibility through stakeholder-relevant examples, while fostering a mindset that includes the entire organization, not just software developers, in the deployment process.It is now time for the automotive industry to transform the academic knowledge and the experiences from other domains to what is applicable for the automotive products and the automotive ecosystem
Scenarios applied to system engineering processes for AD/ADAS safety demonstration
No full textInternational audienceAD/ADAS are complex systems operating in open environments and are subject to high safety expectations. To address the environment diversity, scenarios and use cases are employed to represent possible real-world situations and system's responses in a compressed form. These concepts are integrated throughout the system engineering process to support system design, validation and audit, providing safety assurance evidence. By manipulating these objects in dedicated toolchains, it is possible to systematically manage the combinatorial complexity and demonstrate sufficient coverage of the system's operational conditions
Leveraging the chaotic regime of a MEMS oscillator for gas detection
No full textNational audienceIn harsh environments, measurement noise can severely limit the performance of traditional linear sensors. In this work, we propose exploiting the exponential sensitivity of chaotic systems to minute variations in system parameters to overcome these noise-related limitations1. Specifically, we employ a MEMS Duffing resonator operating in a chaotic regime. Chaos is triggered on demand through a modulated driving frequency2, resulting in a mechanical response in both amplitude and phase that displays chaotic behavior. Importantly, within a limited time window—assuming nearly identical initial conditions—the system dynamics remain reproducible. Within this window, any deviations in the trajectory directly reflect changes in the sensor’s environment, which are then amplified by the system's inherent exponential sensitivity to initial conditions. Crucially, because chaos is deterministic, distinct changes in the initial conditions consistently yield distinct chaotic evolutions, thereby enabling the chaotic patterns to reliably encode information about variations in the physical variable of interest. We demonstrate this concept for the detection of carbon dioxide (CO₂) in nitrogen (N₂): the variation in the gas composition (CO₂:N₂) alters the mixture density, which, in turn, affects the natural resonance frequency of the MEMS device3. Notably, our experiments reveal that the reproducibility time of the chaotic trajectories scales inversely—and monotonously—with the CO₂ concentration, marking a significant milestone for gas sensing using a chaotic regime
Systematic Analysis of the Links Between Obsolescence–Shortage and Reliability–Maintainability–Availability
No full textInternational audienceComponent obsolescence and shortages impact virtually all systems, particularly those with extended lifespans, such as aircrafts or trains. Obsolescence (i.e. state of being outdated) and shortage (i.e. lack of components) are distinct yet interconnected challenges and despite their inevitability, systems must maintain acceptable performance levels over extended periods. In this research, the term 'shortage' encapsulates the broader concepts of Diminishing Manufacturing Sources and Material Shortages (DMSMS). This study investigates the impacts of obsolescence and shortages on maintainability, reliability, and availability. Existing scientific literature frequently overlooks this influence when calculating associated indicators, often relegating it to a marginal factor despite its potential significant impact. The aim is to underscore the criticality of incorporating obsolescence and shortage considerations into operational availability assessments. This argument stems from two key factors: the accelerating pace of obsolescence and shortages driven by rapid technological advancements, evolving needs, logistical disruptions, or global conflicts, and the inherent risks associated with overestimating operational availability. This study encompasses both the system design-development phase and the operational phase. It examines the relevant concepts and their interrelationships through the lens of international standards and established scientific literature. Furthermore, an industrial case study from a major automotive supplier, focusing on one of its production lines, serves to illustrate partially these relationships. The paper concludes by presenting the study's findings and outlining potential avenues for future research in this domain