TIB Open Publishing
Not a member yet
1994 research outputs found
Sort by
AutoML for Industrial Process Control: Analysis of its Benefits and Impact on Real Applications
Aufgrund der zunehmenden Komplexität der modernen Fertigung generieren industrielle Prozesssteuerungssysteme riesige Datenmengen mit erheblichem Potenzial für Anwendungen im Bereich des maschinellen Lernens. Obwohl ML immense Vorteile bietet, stellt der Mangel an Datenwissenschaftskompetenz eine Herausforderung für die Einführung dar. AutoML-Frameworks überwinden diese Hindernisse, indem sie wichtige ML-Aufgaben automatisieren und so die Zugänglichkeit und Effizienz verbessern. Diese Studie untersucht ihre Wirksamkeit in einem Anwendungsfall aus der Keramikindustrie, vergleicht Vorverarbeitungsstrategien und analysiert die Erklärbarkeit mit SHAP-Werten, die von Fachexperten validiert wurden. Die Ergebnisse unterstreichen das Potenzial von AutoML zur Rationalisierung der ML-Modellentwicklung, aber auch seine Abhängigkeit von Fachwissen für eine effektive Merkmalsauswahl und Erklärbarkeit.Due to the growing complexity of modern manufacturing, industrial process control systems generate vast amounts of data with significant potential for machine learning applications. While ML offers immense benefits, the lack of data science expertise poses challenges for adoption. AutoML frameworks tackle these barriers by automating key ML tasks, enhancing accessibility and efficiency. This study investigates their effectiveness in a ceramic industry use case, comparing preprocessing strategies and analyzing explainability with SHAP values validated by domain experts. The findings highlight AutoML\u27s potential to streamline ML model development but also its reliance on domain expertise for effective feature selection and explainability
Alexbek at LLMs4OL 2025 Tasks A, B, and C: Heterogeneous LLM Methods for Ontology Learning (Few-Shot Prompting, Ensemble Typing, and Attention-Based Taxonomies)
We present a comprehensive system for addressing Tasks A, B, and C of the LLMs4OL 2025 challenge, which together span the full ontology construction pipeline: term extraction, typing, and taxonomy discovery. Our approach combines retrieval-augmented prompting, zero-shot classification, and attention-based graph modeling — each tailored to the demands of the respective task.
For Task A, we jointly extract domain-specific terms and their ontological types using a retrieval-augmented generation (RAG) pipeline.Training data was reformulated into a document to terms and types correspondence, while test-time inference leverages semantically similar training examples. This single-pass method requires no model finetuning and improves overall performance through lexical augmentation
Task B, which involves assigning types to given terms, is handled via a dual strategy. In the few-shot setting (for domains with labeled training data), we reuse the RAG scheme with few-shot prompting. In the zero-shot setting (for previously unseen domains), we use a zero-shot classifier that combines cosine similarity scores from multiple embedding models using confidence-based weighting.
In Task C, we model taxonomy discovery as graph inference. Using embeddings of type labels, we train a lightweight cross-attention layer to predict is-a relations by approximating a soft adjacency matrix.
These modular, task-specific solutions enabled us to achieve top-ranking results in the official leaderboard across all three tasks. Taken together these strategies showcase the scalability, adaptability, and robustness of LLM-based architectures for ontology learning across heterogeneous domains.
Code is available at: https://github.com/BelyaevaAlex/LLMs4OL-Challenge-Alexbe
Experimental and Numerical Study of the Thermal Behavior of Clips in Concentrated Solar Tower Receivers
In this study, we investigated the thermal behaviour of a guiding support, referred to as a clip, for a tube resembling those used in solar central receivers. The focus is on the clip\u27s heating process, which occurs when molten salt begins to flow inside the receiver tube. This work involves both experimental measurements and numerical simulations of the clip, which can be considered as a fin attached to the tube. Experimentally, we measured the temperature profile of the clip over time and observed that, despite the characteristic decaying exponential shape of the clip, the heat transfer can be approximated as one-dimensional. Moreover, numerical simulations were performed to gain a better understanding of the clip’s thermal response. By comparing the simulation results to the experimental data, we were able to determine the most suitable boundary condition for the clip end, which is attached to the receiver\u27s supporting structure. Furthermore, the experimental data was compared to an analytical solution for the temperature evolution of a fin that took into account the cooling effect of natural convection and radiation, yielding a maximum error of 8%
NEWS4CSP Project – New Coatings Approaches to Protect Metallic Materials From Heat Transfer Fluids
The efficiency of Concentrated Solar Power technologies increases with the operation temperature of the power block, which can be achieved using advanced power cycles operating at temperatures ≥650°C. Molten Salt (MS) based HTF have been facing critical challenging issues of severe corrosion and lower specific heat capacity. This paper aims to present the NEWS4CSP project to the CSP scientific community, highlighting the main concepts being pursued and the preliminary results obtained regarding the application of new coatings to increase the anticorrosive protection of structural materials in contact with MS. Two different types of powder coatings (nickel and cobalt alloys) were applied to AISI 430 stainless steel using two coating technologies: High Velocity Oxy-Fuel (HVOF) and Laser Cladding (LC). LC proved to be the most suitable technique of these two coating application technologies. The Co-Cr coating alloy showed good anticorrosive behaviour even after 3000 h of testing. The obtained results are very promising, nevertheless, more studies are needed to optimise its application by LC. These include optimizing the thickness and chemical composition, as well as gaining a deeper understanding of the degradation mechanisms and whether this behaviour is maintained over longer periods
Designing an Integrated CSP-SOE System for Hydrogen Production
This paper presents the design, and simulation of a novel concentrated solar power (CSP) system integration with solid oxide electrolysis (SOE) to generate superheated steam required for efficient hydrogen production. The system comprises of 10 parabolic dish collectors, SiSiC cavity receivers, and a heat exchanger with two components. An evaporator and a superheater collectively achieve a high thermal efficiency of 73%. Computational fluid dynamics (CFD) simulations demonstrated that the SiSiC receiver can maintain an air outlet temperature of 1555K, which is 8% more efficient than a similar design reported in the literature. This would facilitate the generation of superheated steam necessary for the SOE process. A water flow rate of 28.8kg/h directly influences the system\u27s hydrogen production capacity, which reaches 2.56kg/h at optimal conditions. The heat exchanger components were designed using Aspen EDR, while the entire system simulation was conducted using Aspen Plus, demonstrating the system\u27s potential to meet industrial standards for sustainable hydrogen production. This article serves as a good reference in investigating the feasibility of CSP-SOE systems as a promising pathway for large-scale renewable hydrogen production
Review of the Operational Phase of the South African Concentrated Solar Power Projects: Overall Performance to Date
This paper provides an overview of the six Concentrated Solar Power (CSP) projects in operation in South Africa, comprising 500 MW generation capacity. These projects, in addition to the seventh (100 MW) CSP project currently in construction, were awarded under various bid windows of the Renewable Energy Independent Power Producer’s Procurement Programme (REIPPPP) and commenced operations during 2015-2019. Key operational data up to the end of 2023 are presented, as of when the combined cumulative electricity generation by the six CSP projects reached 10.1 TWh. This represents approximately 81% of the sum of the P50 Energy Yield Forecasts (EYFs) in the respective project Power Purchase Agreements (PPAs), and 9.6% of all electricity generated by the REIPPPP projects by the end of 2023, while the CSP projects constitute 8.1% of total installed capacity (6180 MW) as of December 2023. Factors affecting performance are discussed, including meteorology and outages. Consumption of major utilities is also discussed, comparing to expected values or limits
JCA Eni ENEA Project: CSP & Thermal Storage
This project is framed inside a three-year-long cooperation between Eni and ENEA. It concerns the development of prototypes, technologies, innovative processes, feasibility studies, scenario analysis, exchanging skills, and know-how. This is coherent with the promotion of joint initiatives in the field of energy production, from renewable sources and those with low CO2 emissions, supercomputing, superconductivity, and circular economy, including innovative processes for waste valorisation and production of biogas, biomethane and biochar. The “CSP & Thermal Storage” has the ambition to exploit solar sources to produce thermal energy according to the energy demand, regardless of the availability of the solar source, through the development of innovative solutions of plants with a programmable production capacity, through the coupling with innovative Thermal Energy Storage (TES) solutions. The objectives of this project are the identification of contexts for using Concentrating Solar Power / Concentrating Solar Heat (CSP/CSH) technology and thermal generation systems coupled to TES, the definition of the "business plan" for the enhancement of these technologies, the identification of partnerships with national and international subjects for the design of industrial solutions in the identified contexts of interest, and the joint participation in competitive tenders through research and development projects that can also lead to the creation, characterisation, and validation of prototype units. As for the energy storage, dynamic modelling of the thermocline and Phase Change Materials (PCM) storage systems is also expected, together with a broad-spectrum analysis of possible applications of CSP/CSH coupled with TES along the entire industrial energy supply chain (e.g.: upstream, downstream, power generation). Of particular importance is the development of TES prototypes to be tested on a molten salt-operated circuit integrating two different PCM systems with an Eni proprietary innovative thermocline TES based on concrete, to facilitate the management of the system and to supply high-quality heat to the user. This system consists of a module able to store 40 kWh of thermal energy by phase change materials (PCM), followed by a concrete module of about 150 kWh and another 40 kWh PCM TES with a higher phase change temperature. The overall operating temperature range of the system is 290÷450°C
Summary of Standardization for Concentrating Solar Plants in Industrial Processes
The Solar Heat Integration in Industrial Processes (SHIP) covers a wide range of temperature applications, from low temperature for which need flat plate collectors (FPC) or evacuated tubes collectors (ETC), to high temperature which need larger parabolic-trough collectors (PTC), linear Fresnel (LFR) or solar tower (ST), like the one installed in Concentrating Solar Power (CSP) plants. This paper aims to present a summary of all the existing standards (published or in work process) that should be used in a SHIP plant to regulate and ensure the products used in those renewable energy thermal plants
Design of A Winch-Actuated Heliostat: SolarPACES
This study addresses the challenge of reducing costs in concentrated solar power (CSP) systems by developing a novel winch-actuated heliostat design. Building upon previous cable-actuated designs, the proposed heliostat utilises steel cables attached to the reflective surface frame, with a worm gear set to decrease torque demands. Through computer-aided design, prototype development, and wind load calculations, the research demonstrates the potential of this design to significantly reduce heliostat costs while maintaining the required tracking accuracy of 1 mrad. This innovation contributes to efforts to achieve the U.S. Department of Energy\u27s 2030 cost target of $ 50/m² for solar fields, potentially advancing the competitiveness of CSP technology
Opto-Electronic Sensor for HTF Early Leak Detection Based on Infrared Absorption
We propose an early leak detection sensor for HTF (Heat Transfer Fluid) leakages by measuring the absorption of infrared radiation by the HTF components. This sensor, that can be installed inside the insulation around ball joints or other elements in a Concentrated Solar Power (CSP) plant, detects leaks by sensing mid-infrared absorption at specific wavelengths. The sensor includes a reference signal to correct for external factors like temperature changes and sensor degradation. Laboratory tests have validated the sensor performance under various conditions. The sensor enhances the sustainability of solar thermal plants by providing early leak detection and is easily adjustable for different HTFs by changing the IR (Infrared) radiation sensing bands