Filtern
Erscheinungsjahr
Dokumenttyp
- Vortrag (19)
- Zeitschriftenartikel (16)
- Beitrag zu einem Tagungsband (3)
- Video (2)
- Posterpräsentation (2)
- Forschungsdatensatz (2)
- Sonstiges (1)
Sprache
- Englisch (37)
- Deutsch (7)
- Mehrsprachig (1)
Schlagworte
- Concrete (9)
- Data fusion (9)
- Machine learning (9)
- Machine Learning (8)
- NDT (7)
- Corrosion (5)
- Clustering (4)
- Concrete evaluation (4)
- Honeycombing (4)
- Materials Informatics (4)
Organisationseinheit der BAM
- 8 Zerstörungsfreie Prüfung (33)
- 8.0 Abteilungsleitung und andere (32)
- 7 Bauwerkssicherheit (3)
- 5 Werkstofftechnik (2)
- 5.2 Metallische Hochtemperaturwerkstoffe (2)
- 7.6 Korrosion und Korrosionsschutz (2)
- 1 Analytische Chemie; Referenzmaterialien (1)
- 1.3 Instrumentelle Analytik (1)
- 5.1 Mikrostruktur Design und Degradation (1)
- 7.7 Modellierung und Simulation (1)
Paper des Monats
- ja (1)
Eingeladener Vortrag
- nein (19)
Das maschinelle Lernen (ML) wurde erfolgreich zur Lösung vieler Aufgaben in der zerstörungsfreien Prüfung im Bauwesen (ZfPBau) eingesetzt. Allerdings ist die Erstellung von Referenzdaten in den meisten Fällen extrem teuer und daher viel knapper als in anderen Forschungsbereichen. Auch decken die verfügbaren Daten mitunter nur ein einziges Szenario ab, so dass die Leistungsindikatoren oft nicht die tatsächliche Leistung des ML-Modells in der praktischen Anwendung widerspiegeln. Schätzungen, die die Übertragbarkeit von einem Szenario auf ein anderes quantifizieren, sind erforderlich, um dieser Herausforderung gerecht zu werden und den Weg für Anwendungen in der Praxis zu ebnen.
In diesem Beitrag stellen wir Werkzeuge zur Beschreibung der Unsicherheit von ML in neuen ZfPBau-Szenarien vor. Zu diesem Zweck haben wir einen bestehenden Trainingsdatensatz zur Klassifizierung von Korrosionsschäden der Bewehrung in Beton um eine neue Fallstudie erweitert. Die Messungen wurden an großflächigen Betonproben mit eingebauter chloridinduzierter Korrosion des Bewehrungsstahls durchgeführt. Das Experiment simulierte den gesamten Lebenszyklus von chloridinduzierten Sichtbetonbauteilen im Labor. Unser Datensatz umfasst Potenzialfeld- und Radarmessungen. Die einzigartige Fähigkeit, die Schädigung zu überwachen und eine gezielte Korrosion einzuleiten, ermöglichte es, die Daten zu labeln - was für die Konstruktion von ML-Modellen entscheidend ist. Um die Übertragbarkeit zu untersuchen, erweitern wir unser Modell um Metadaten - wie etwa Konstruktionsmerkmale des Prüfkörpers und Umweltbedingungen. Dies erlaubt es, die Veränderung dieser Merkmale in neuen Szenarien mit statistischen Methoden als Unsicherheiten auszudrücken. Wir vergleichen verschiedene auf Stichproben und statistischer Verteilung basierende Ansätze und zeigen, wie diese Methoden eingesetzt werden können, um Wissenslücken von ML-Modellen in der ZfP zu schließen
ML has been successfully applied to solve many NDT-CE tasks. This is usually demonstrated with performance metrics that evaluate the model as a whole based on a given set of data. However, since in most cases the creation of reference data is extremely expensive, the data used is generally much sparser than in other areas, such as e-commerce. As a result, performance indicators often do not reflect the practical applicability of the ML model. Estimates that quantify transferability from one case to another are necessary to meet this challenge and pave the way for real world applications.
In this contribution we invetigate the uncertainty of ML in new NDT-CE scenarios. For this purpose, we have extended an existing training data set for the classification of corrosion damage by a new case study. Our data set includes half-cell potential mapping and ground-penetrating radar measurements. The measurements were performed on large-area concrete samples with built-in chloride-induced corrosion of reinforcement. The experiment simulated the entire life cycle of chloride induced exposed concrete components in the laboratory. The unique ability to monitor deterioration and initiate targeted corrosion initiation allowed the data to be labelled - which is crucial to ML. To investigate transferability, we extend our data by including new design features of the test specimen and environmental conditions. This allows to express the change of these features in new scenarios as uncertainties using statistical methods. We compare different sampling and statistical distribution-based approaches and show how these methods can be used to close knowledge gaps of ML models in NDT.
Environmentally friendly alternatives to cement are created through the synthesis of numerous base materials. The variation of their proportions alone leads to millions of materials candidates. Identifying suitable materials is very laborious; traditional systematic research in the laboratory consumes a lot of time and effort.
Sequential learning (SL) potentially speeds up the materials research process despite limited but highly complex available information. SL does not make direct predictions of material properties but ranks possible experiments according to their utility. The most promising experiments are prioritized over dead-end experiments and experiments whose outcome is already known.
Our work has shown that SL seems to be promising for cement research. So far, research has mainly focused on materials whose synthesis is faster and whose material properties require less time for development or characterization (allowing many successive experiments). Contrarily, in the case of binders, SL is only useful if few experiments lead to the desired goal, as for example, the determination of the compressive strength alone typically requires 28 days.
In research practice, experimental designs and the availability of resources often determine which data can be used - for example, when some laboratory resources are not available or deemed irrelevant to a task. As a result, new research scenarios are constantly emerging, each of which requires to demonstrate SL’s performance.
We are presenting the SLAMD app to facilitate the exploration of SL methods in numerous research scenarios. The app provides flexible and low-threshold access to AI methods via intuitive and interactive user interfaces. We deliberately pursue a software-based research approach (as opposed to code-, or script-based). On the one hand, the results are more comprehensible since we refer to a common (code) basis (’reproducible science’). On the other hand, the methods are easily accessible to all which accelerates the knowledge transfer into laboratory practice.
An Ontology-Based Approach to Enable Data-Driven Research in the Field of NDT in Civil Engineering
(2021)
Although measurement data from the civil engineering sector are an important basis for scientific analyses in the field of non-destructive testing (NDT), there is still no uniform representation of these data. An analysis of data sets across different test objects or test types is therefore associated with a high manual effort. Ontologies and the semantic web are technologies already used in numerous intelligent systems such as material cyberinfrastructures or research databases. This contribution demonstrates the application of these technologies to the case of the 1H nuclear magnetic resonance relaxometry, which is commonly used to characterize water content and porosity distri-bution in solids. The methodology implemented for this purpose was developed specifically to be applied to materials science (MS) tests. The aim of this paper is to analyze such a methodology from the perspective of data interoperability using ontologies. Three benefits are expected from this ap-proach to the study of the implementation of interoperability in the NDT domain: First, expanding knowledge of how the intrinsic characteristics of the NDT domain determine the application of semantic technologies. Second, to determine which aspects of such an implementation can be improved and in what ways. Finally, the baselines of future research in the field of data integration for NDT are drawn.
This paper studies the dispersion of GPR wave’s phase velocity at different wideband frequencies in plywood and concrete with varying moisture content. This study makes use of two GPR antennas with 2 GHz centre frequency operating in wide angle reflection and refraction (WARR) mode and with computation of spectral analysis of the surface wave (SASW). Computation of phase velocities is based on the Acquisition of the cross-power spectrum and phase unwrap of two distorted ground waves at positions Closer to and farther away from the transmitting antenna. The velocities of the ground waves are found to experience greater dispersion in low frequency regimes within the effective frequency bandwidths determined and thresholded by time–frequency analysis (TFA) and coherence plotting of the ground waves.
This study validates not only the methodology, but also identifies the optimal distance between the first (Rx1) and second (Rx2) receivers as k/2, which is based on a fixed transmitter (Tx) minus the first Receiver (Rx1) distance. It serves as an indication of changeable separation distance when other lower frequency GPR is used because the distances of Tx-Rx1 and Rx1-Rx2 are wavelength dependent and thus also frequency dependent. Effects of moisture contents and chloride contamination in concrete were also characterized according to the dispersion plots . that wave traelling in lower frequencies in GPR wave is much decelerated than that in higher frequencies. This research also contributes to the building of the ‘‘GPRWARR machine” suggested in Annan and Jackson (2017) [1], within which the effects of wave Dispersion on phase velocity can be inversely modelled to characterize variations in the material properties of infrastructure as a means of detecting surface damage.
Alkali-activated binders (AAB) can provide a clean alternative to conventional cement in terms of CO2 emissions. However, as yet there are no sufficiently accurate material models to effectively predict the AAB properties, thus making optimal mix design highly costly and reducing the attractiveness of such binders. This work adopts sequential learning (SL) in high-dimensional material spaces (consisting of composition and processing data) to find AABs that exhibit desired properties. The SL approach combines machine learning models and feedback from real experiments. For this purpose, 131 data points were collected from different publications. The data sources are described in detail, and the differences between the binders are discussed. The sought-after target property is the compressive strength of the binders after 28 days. The success is benchmarked in terms of the number of experiments required to find materials with the desired strength. The influence of some constraints was systematically analyzed, e.g., the possibility to parallelize the experiments, the influence of the chosen algorithm and the size of the training data set. The results show the advantage of SL, i.e., the amount of data required can potentially be reduced by at least one order of magnitude compared to traditional machine learning models, while at the same time exploiting highly complex information. This brings applications in laboratory practice within reach.
In our current research project „Reincarnate“ we aim to anchor the idea of the circular economy in the European construction industry and significantly extend the life cycle of buildings, construction products and materials through innovative solutions. On the long term, this is an approach reduce construction waste by 80 percent and the CO2 footprint of the construction sector by 70 percent."
This project has received funding from the European Union’s Horizon Europe research and innovation programme and will take you on a tour on what are the drivers, what is the goal, who are the partners and how we want to make the world a better place!
With 8% of man-made CO2 emissions, cement production is an important driver of the climate crisis. By using alkali-activated binders, part of the energy-intensive clinker production process can be dispensed. However, as numerous raw materials are involved in the manufacturing process here, the complexity of the materials increases by orders of magnitude. Finding a properly balanced binder formulation is like looking for a needle in a haystack. We have shown for the first time that artificial intelligence (AI)-based optimization of alkali-activated binder formulations can significantly accelerate research.
The "Sequential Learning App for Materials Discovery" (SLAMD) aims to accelerate practice transfer. With SLAMD, materials scientists have low-threshold access to AI through interactive and intuitive user interfaces. The value added by AI can be determined directly. For example, the CO2 emissions saved per ton of cement can be determined for each development cycle: the more efficient the AI optimization, the greater the savings.
Our material database already includes more than 120,000 data points of alternative binders and is constantly being expanded with new parameters. We are currently driving the enrichment of the data with a life cycle analysis of the building materials.
Based on a case study we show how intuitive access to AI can drive the adoption of techniques that make a real contribution to the development of resource-efficient and sustainable building materials of the future and make it easy to identify when classical experiments are more efficient.
With 8% of man-made CO2 emissions, cement production is an important driver of the climate crisis. By using alkali-activated binders, part of the energy-intensive clinker production process can be dispensed. However, as numerous raw materials are involved in the manufacturing process here, the complexity of the materials increases by orders of magnitude. Finding a properly balanced binder formulation is like looking for a needle in a haystack. We have shown for the first time that artificial intelligence (AI)-based optimization of alkali-activated binder formulations can significantly accelerate research.
The "Sequential Learning App for Materials Discovery" (SLAMD) aims to accelerate practice transfer. With SLAMD, materials scientists have low-threshold access to AI through interactive and intuitive user interfaces. The value added by AI can be determined directly. For example, the CO2 emissions saved per ton of cement can be determined for each development cycle: the more efficient the AI optimization, the greater the savings.
Our material database already includes more than 120,000 data points of alternative binders and is constantly being expanded with new parameters. We are currently driving the enrichment of the data with a life cycle analysis of the building materials.
Based on a case study we show how intuitive access to AI can drive the adoption of techniques that make a real contribution to the development of resource-efficient and sustainable building materials of the future and make it easy to identify when classical experiments are more efficient.
SLAMD-FIB-Case-Study
(2022)
With 8% of man-made CO2 emissions, cement production is an important driver of the climate crisis. By using alkali-activated binders part of the energy-intensive clinker production process can be dispensed with. However, because numerous chemicals are involved in the manufacturing process here, the complexity of the materials increases by orders of magnitude. Finding a properly balanced cement formulation is like looking for a needle in a haystack. We have shown for the first time that artificial intelligence (AI)-based optimization of cement formulations can significantly accelerate research. The „Sequential Learning App for Materials Discovery“ (SLAMD) aims to accelerate practice transfer. With SLAMD, materials scientists have low-threshold access to AI through interactive and intuitive user interfaces. The value added by AI can be determined directly. For example, the CO2 emissions saved per ton of cement can be determined for each development cycle: the more efficient the AI optimization, the greater the savings. Our material database already includes more than 120,000 data points of alternative cements and is constantly being expanded with new parameters. We are currently driving the enrichment of the data with a life cycle analysis of the building materials. Based on a case study we show how intuitive access to AI can drive the adoption of techniques that make a real contribution to the development of resource-efficient and sustainable building materials of the future and make it easy to identify when classical experiments are more efficient.
High greenhouse gas emissions from the production of building materials are a major contributor to the current climate crisis. However, developing alternative building materials is complex. Traditional laboratory methods are reaching their limits. Artificial intelligence, on the other hand, can give research a new dynamic.
Novel materials are usually developed manually in the laboratory rather than on a computer. This makes the processes time-consuming, difficult and expensive. With the app SLAMD (Sequential Learning App for Materials Discovery), materials researchers can explore the potential of artificial intelligence to speed up materials research and easily apply AI in the lab. The app was developed by our team at the Federal Institute for Materials Research and Testing (BAM) led by Prof. Sabine Kruschwitz together with a team in the Department of Building Materials and Construction Chemistry at TU Berlin led by Prof. Dietmar Stephan.
It uses material composition and characterization data to predict ideal material candidates. It can be used to optimize many material properties simultaneously and even incorporates database information such as carbon footprint, material cost or resource availability. Unlike the usual data-intensive AI methods, SLAMD optimally integrates existing knowledge and human feedback, and provides numerous decision support tools to precisely navigate complex scientific knowledge processes towards success.
In this talk, we will present some case studies where we were able to find suitable advanced materials in a few months instead of several years. We will talk about the challenges we overcame and the future potential we see for this approach to developing the green materials of the future.
With 8% of man-made CO2 emissions, cement production is an important driver of the climate crisis. By using alkali-activated binders, part of the energy-intensive clinker production process can be dispensed. However, as numerous raw materials are involved in the manufacturing process here, the complexity of the materials increases by orders of magnitude. Finding a properly balanced binder formulation is like looking for a needle in a haystack. We have shown for the first time that artificial intelligence (AI)-based optimization of alkali-activated binder formulations can significantly accelerate research.
The "Sequential Learning App for Materials Discovery" (SLAMD) aims to accelerate practice transfer. With SLAMD, materials scientists have low-threshold access to AI through interactive and intuitive user interfaces. The value added by AI can be determined directly. For example, the CO2 emissions saved per ton of cement can be determined for each development cycle: the more efficient the AI optimization, the greater the savings.
Our material database already includes more than 120,000 data points of alternative binders and is constantly being expanded with new parameters. We are currently driving the enrichment of the data with a life cycle analysis of the building materials.
Based on a case study we show how intuitive access to AI can drive the adoption of techniques that make a real contribution to the development of resource-efficient and sustainable building materials of the future and make it easy to identify when classical experiments are more efficient.
WEBSLAMD
(2022)
The objective of SLAMD is to accelerate materials research in the wet lab through AI. Currently, the focus is on sustainable concrete and binder formulations, but it can be extended to other material classes in the future.
1. Summary
Leverage the Digital Lab and AI optimization to discover exciting new materials Represent resources and processes and their socio-economic impact.
Calculate complex compositions and enrich them with detailed material knowledge. Integrate laboratory data and apply it to novel formulations. Tailor materials to the purpose to achieve the best solution.
Workflow
Digital Lab
Specify resources: From base materials to manufacturing processes – "Base" enables a detailed and consistent description of existing resources
Combine resources: The combination of base materials and processes offers an almost infinite optimization potential. "Blend" makes it easier to design complex configurations.
Digital Formulations: With "Formulations" you can effortlessly convert your resources into the entire spectrum of possible concrete formulations. This automatically generates a detailed set of data for AI optimization.
AI-Optimization
Materials Discovery: Integrate data from the "Digital Lab" or upload your own material data. Enrich the data with lab results and adopt the knowledge to new recipes via artificial intelligence. Leverage socio-economic metrics to identify recipes tailored to your requirements.
We present the results of a machine learning (ML)- inspired data fusion approach, applied to multi-sensory nondestructive testing (NDT) data. Our dataset consists of Impact-Echo (IE), Ultrasonic Pulse Echo (US) and Ground Penetrating Radar (GPR) measurements collected on large-scale concrete specimens with built–in simulated honeycombing defects. In a previous study we were able to improve the detectability of honeycombs by fusing the information from the three different sensors with the density based clustering algorithm DBSCAN. We demonstrated the advantage of data fusion in reducing the false positives up to 10% compared to the best single sensor, thus, improving the detectability of the defects. The main objective of this contribution is to investigate the generality, i.e. whether the conclusions from one specimen can be adapted to the other. The effectiveness of the proposed approach on a separate full-scale concrete specimen was evaluated.
We present the results of a machine learning (ML)- inspired data fusion approach, applied to multi-sensory nondestructive testing (NDT) data. Our dataset consists of Impact-Echo (IE), Ultrasonic Pulse Echo (US) and Ground Penetrating Radar (GPR) measurements collected on large-scale concrete specimens with built–in simulated honeycombing defects. In a previous study we were able to improve the detectability of honeycombs by fusing the information from the three different sensors with the density based clustering algorithm DBSCAN. We demonstrated the advantage of data fusion in reducing the false positives up to 10% compared to the best single sensor, thus, improving the detectability of the defects. The main objective of this contribution is to investigate the generality, i.e. whether the conclusions from one specimen can be adapted to the other. The effectiveness of the proposed approach on a separate full-scale concrete specimen was evaluated.
For the closure of radioactive waste disposal facilities engineered barriers- so called “drift seals” are used. The purpose of these barriers is to constrain the possible infiltration of brine and to prevent the migration of radionuclides into the biosphere. In a rock salt mine a large scale in-situ experiment of a sealing construction made of salt concrete was set up to prove the technical feasibility and operability of such barriers. In order to investigate the integrity of this structure, non-destructive ultrasonic measurements were carried out.
Therefore two different methods were applied at the front side of the test-barrier:
1 Reflection measurements from boreholes
2 Ultrasonic imaging by means of scanning ultrasonic echo methods This extended abstract is a short version of an article to be published in a special edition of ASCE Journal that will briefly describe the sealing construction, the application of the non-destructive ultrasonic measurement methods and their adaptation to the onsite conditions -as well as parts of the obtained results. From this a concept for the systematic investigation of possible contribution of ultrasonic methods for quality assurance of sealing structures may be deduced.
We present the results of several machine learning (ML)- inspired data fusion algorithms applied to multi-sensory nondestructive testing (NDT) data. Our dataset consists of Impact-Echo (IE), Ultrasonic Pulse Echo (US) and Ground Penetrating Radar (GPR) data collected on large-scale concrete specimens with built–in simulated honeycombing defects. The main objective is to improve the detectability of honeycombs by fusing the information from the three different sensors. We describe normalization, feature detection and optimal feature selection. We have used unsupervised and supervised ML, i.e., classification and clustering, for data fusion. We demonstrate the advantage of data fusion in reducing the false positives up to 10% compared to the best single sensor, thus, improving the detectability of the defects. The methods were evaluated on a concrete specimen. The effectiveness of the proposed approach was demonstrated on a separate full-scale concrete specimen. The results indicate the transportability of the conclusions from one specimen to the other.
Potentialfeldmessung (PM) ist die beliebteste Methode der Zerstörungsfreien Prüfung (ZfP) zur Lokalisierung von aktiver Betonstahlkorrosion. PM wird durch Parameter wie z. B. Feuchtigkeits- und Chloridgradienten im Bauteil beeinflusst, so dass die Sensitivität gegenüber der räumlich sehr begrenzten, aber gefährlichen Lochkorrosion gering ist. Wir zeigen in dieser Studie, wie zusätzliche Messinformationen mit Multisensor-Datenfusion genutzt werden können, um die Detektionsleistung zu verbessern und die Auswertung zu automatisieren. Die Fusion basiert auf überwachtem maschinellen Lernen (ÜML). ÜML sind Methoden, die Zusammenhänge in (Sensor-) Daten anhand vorgegebener Kennzeichnungen (Label) erkennen. Wir verwenden ÜML um „defekt“ und „intakt“ gelabelte Bereiche in einem Multisensordatensatz zu unterscheiden. Unser Datensatz besteht aus 18 Messkampagnen und enthält jeweils PM-, Bodenradar-, Mikrowellen-Feuchte- und Wenner-Widerstandsdaten. Exakte Label für veränderliche Umweltbedingungen wurden in einer Versuchsanordnung bestimmt, bei der eine Stahlbetonplatte im Labor kontrolliert und beschleunigt verwittert. Der Verwitterungsfortschritt wurde kontinuierlich überwacht und die Korrosion gezielt erzeugt. Die Detektionsergebnisse werden quantifiziert und statistisch ausgewertet. Die Datenfusion zeigt gegenüber dem besten Einzelverfahren (PM) eine deutliche Verbesserung. Wir beschreiben die Herausforderungen datengesteuerter Ansätze in der zerstörungsfreien Prüfung und zeigen mögliche Lösungsansätze.
This talk demonstrates the results of the IGSTC-project entitled "NDT-Data Fusion".
Project approach:
Nondestructive testing (NDT) of concrete buildings allows to coordinate efficient repair measures. Multi-sensor platforms collect large data sets. Nevertheless, data analysis is typically performed manually. Data Fusion uses the full potential of a multi-sensory data set in order to:
improve information quality (reliability, robustness, accuracy, clarity, completeness) and enables automated algorithm based data analysis.
We present the project achievements, namely:
- Development of building scanner system for multisensory NDT
- Laboratory multi sensor investigations
- Development of data fusion concept for honeycombing and pitting corrosion
- Field testing
Half-cell potential mapping (HP) is the most popular nondestructive test (NDT)-method for the localization of corrosion damage in concrete. It is generally recognized, that HP is prone to the environmental factors that arise from salt induced deterioration, such as varying moisture and chloride gradients. Additional NDT-methods are capable to determine distinctive areas, but cannot yet be used to estimate more accurate testing results. We introduce a supervised machine learning (SML) based approach for data fusion to make use of the additional sensor information. SML are methods that explore relations between different (sensor) data from predefined data labels. We use a simple linear classifier named logistic regression to distinguish defect and intact areas. The test performance improves drastically compared to the best single method, HP. In order to generate representative, labeled data we conducted a comprehensive experiment that simulates the deterioration-cycle of a chloride-exposed building part in the lab. Our data set consist of 18 measurement campaigns, each containing HP-, ground-penetrating-radar-, microwave-moisture-, and Wenner-resistivity-data. We detail the challenges that arise with a data driven approach in NDT and how we addressed them.
Research and Development in Non-Destructive Testing in Civil Engineering at BAM has de-veloped over more than two decades. Highlights of present research are ultrasound phase array probes, ground penetrating radar applications for the assessment of existing structures and Laser Induced Breakdown Spec-troscopy (LIBS) to determine which chemical elements on the surface of a specimen are. RFID sensor devel-opments for wireless moisture monitoring and data fusion for the combined analysis of sensor data are fast developing areas. In addition, a selection of application cases is briefly described.
Angetrieben durch Digitalisierung und die sogenannte Industrie 4.0 steigt die Erwartungshaltung gegenüber der Nutzung von Sensorik. Die Unsicherheit, Widersprüchlichkeit und Redundanz separat verarbeiteter einzelner Quellen soll durch die synergetische Zusammenführung heterogener Datensätze überwunden werden. Dabei steht der zunehmenden Aufgabenkomplexität eine ebenso zunehmende Verfügbarkeit an Sensoren, Daten und Rechenleistung gegenüber.
Der Begriff Datenfusion fasst Ansätze zusammen, die Daten zu abstrakteren, aber besser verständlichen Informationen verarbeiten. Die Fusion ist Kernbestandteil effektiver Assistenzsysteme und beweist in vielzähligen Aufgaben - von militärischen Anwendungen über Flug- und Fahrassistenzsysteme, bis in den Heimbereich – ihr großes Potenzial. Durch die wachsende Automatisierung bei der Messdatenerfassung wird Datenfusion auch in der industriellen Qualitätsprüfung und –sicherung zunehmend attraktiver.
Der Vortrag gibt einen Überblick über den breiten Themenkomplex und widmet sich dabei im Theorieteil speziell der Fragen, welche Informationen in multivariaten Datensätzen stecken und wie sie extrahiert werden können. Anschließend wird ein Beispiel für die erfolgreiche Anwendung zur zerstörungsfreien Prüfung von Betonbauteilen vorgestellt. Der dargestellte Datensatz ist klein, heterogen, hochdimensional und unausgeglichen. Anhand von Algorithmen mit unterschiedlicher Leistungsfähigkeiten hinsichtlich Anpassungsfähigkeit und Invarianz gegenüber Höherdimensionalität wird erläutert welche Prozesse zur Verbesserung der Informationsqualität nötig sind.
This work presents machine learning-inspired data fusion approaches to improve the non-destructive testing of reinforced concrete. The principal effects that are used for data fusion are shown theoretically. Their effectiveness is tested in case studies carried out on largescale concrete specimens with built-in chloride-induced rebar corrosion. The dataset consists of half-cell potential mapping, Wenner resistivity, microwave moisture and ground penetrating radar measurements. Data fusion is based on the logistic Regression algorithm.
It learns an optimal linear decision boundary from multivariate labeled training data, to separate intact and defect areas. The training data are generated in an experiment that simulates the entire life cycle of chloride-exposed concrete building parts. The unique possibility to monitor the deterioration, and targeted corrosion initiation, allows data labeling.
The results exhibit an improved sensitivity of the data fusion with logistic regression compared to the best individual method half-cell potential.
Explore and Exploit - Strategische Erweiterung der fraktographischen Datenbank mit Machine Learning
(2020)
In diesem Vortrag stellen wir den aktuellen Stand zu einer Masterarbeit zusammen die sich mit dem Thema beschäftigt wie die Generalisierbarkeit von Datenmodellen auf Basis kleiner Datensätze erhöht werden kann. Wir stellen vor, wie die Datenbasis eines fraktogafischen Bildklassifizierers mit einem statistischen Model strategische erweitert, bzw. an eine Anwendung angepasst werden kann.
We have arrived in the data age. But why is it so difficult for the NDT community to achieve real breakthroughs with data-driven science? In this seminar, we will take a brief look at the evolution of mainstream data science to understand why the most exciting times are perhaps just ahead. We will give an overview of our activities in the junior research group 8.K which are aimed at enabling the next generation of data science methods in NDT. The seminar addresses the two main work fields of our group: semantic data management and the handling of limited data resources.
The first field addresses the problem that a uniform representation of our data is not yet available. However, knowledge creation in data science - whose main contribution lies in the analysis of distributed resources - requires common data access based on a collective understanding. To achieve this, we present an ontology-based approach. Ontologies are already the core of many intelligent systems such as building information models or research databases. We summarize some of the basic principles of this technology and describe our approach to create an NDT ontology.
The second field ties in with the first and addresses the application of data-based methods in engineering practice. Especially in the field of non-destructive testing many successful applications have been published. In most cases, however, the creation of referenced data is extremely expensive and therefore much sparser than in other research areas. As a result, the available data may cover only one scenario, so that common benchmarks often do not reflect the actual performance of the model in practical applications. Estimates that quantify the transferability from one scenario to another are not only necessary to overcome this challenge - they also prove to be a powerful tool for the strategic expansion of what we consider knowledge.
Data-driven research is considered the new paradigm in science. In this field, data is the new resource from which knowledge is extracted that is too complex for traditional methods. Several factors such as national funding and advances in information technology, are driving the development. In particular, the creation of databases and the analysis of data with artifical intelligence are playing an important role in establishing the new paradigm. However, there are numerous challenges that must be overcome to realize the full potential of data-driven methods. This talk sets the stage for the upcoming workshop by reviewing some of the historical developments and the current state of data-driven science in NDT and materials science.
Many PhD students are interested in applying machine learning, AI, data science, etc., and there are many good reasons for this. However, there is a disconnect between mainstream data science and materials science, for example, when it comes to the sheer size of the data. This talk will highlight some of the unique challenges in materials informatics and present some interesting approaches to overcome them. Although the field is large, this talk will focus on cases that have some practical relevance to PhD students at BAM.
In the field of non-destructive testing (NDT) in civil engineering, a large number of measurement data are collected. Although they serve as a basis for scientific analyses, there is still no uniform representation of the data. An analysis of various distributed data sets across different test objects is therefore only possible with high manual effort.
We present a system architecture for an integrated data management of distributed data sets based on Semantic Web technologies. The approach is essentially based on a mathematical model - the so-called ontology - which represents the knowledge of our domain NDT. The ontology developed by us is linked to data sources and thus describes the semantic meaning of the data. Furthermore, the ontology acts as a central concept for database access. Non-domain data sources can be easily integrated by linking them to the NDT construction ontology and are directly available for generic use in the sense of digitization. Based on an extensive literature research, we outline the possibilities that this offers for NDT in civil engineering, such as computer-aided sorting, analysis, recognition and explanation of relationships (explainable AI) for several million measurement data.
The expected benefits of this approach of knowledge representation and data access for the NDT community are an expansion of knowledge through data exchange in research (interoperability), the scientific exploitation of large existing data sources with data-based methods (such as image recognition, measurement uncertainty calculations, factor analysis, material characterization) and finally a simplified exchange of NDT data with engineering models and thus with the construction industry.
Ontologies are already the core of numerous intelligent systems such as building information modeling or research databases. This contribution gives an overview of the range of tools we are currently creating to communicate with them.
Im Bereich der Zerstörungsfreien Prüfung (ZfP) im Bauwesen werden eine Vielzahl von Messdaten erfasst. Obwohl Sie als Grundlage für wissenschaftliche Analysen dienen, gibt es noch keine einheitliche Repräsentation der Daten. Eine Analyse verschiedener verteilter Datensätze über unterschiedliche Prüfobjekte hinweg ist daher kaum möglich.
Wir stellen einen Ansatz für ein integriertes Datenmanagement verteilter Datensätze auf Basis von Semantic-Web Technologien vor. Der Ansatz basiert im Kern auf einem mathematischen Modell – der sogenannten Ontologie – welches das Wissen unserer Domäne ZfPBau repräsentiert. Die von uns entwickelte ZfPBau Ontologie wird mit Datenquellen verknüpft und beschreibt so die semantische Bedeutung der Daten. Darüber hinaus fungiert die Ontologie als zentrales Konzept für den Datenbankzugriff. Domänen-fremde Datenquellen können durch die Verknüpfung mit der ZfPBau Ontologie einfach integriert werden und stehen zur generischen Nutzung im Sinne der Digitalisierung direkt zur Verfügung. Basierend auf einer umfangreichen Literaturrecherche, skizzieren wir die Möglichkeiten die sich daraus für die ZfP im Bauwesen ergeben, wie zum Beispiel Messdaten computergestützt zu sortieren, zu analysieren, Zusammenhänge zu erkennen und zu erklären.
Der erwartete Nutzen dieses Ansatzes der Wissensrepräsentation und des Datenzugriffs für die ZfP-Community ist eine Erweiterung des Wissens durch Datenaustausch in der Forschung (Interoperabilität), die wissenschaftliche Verwertung großer existierender Datenquellen mit datenbasierten Verfahren (wie Bilderkennung, Messunsicherheitsberechnungen, Faktoranaylsen, Materialcharackterisierung) und letztlich ein vereinfachter Transfer von ZFP-Daten in Ingenieurmodelle und somit in die Baupraxis.
Ontologien sind bereits Kern vielzähliger intelligenter Systeme wie Building-Information-Modeling oder Forschungsdatenbanken. Der Beitrag gibt einen Überblick über die Werkzeuge die wir derzeit für die Kommunikation mit ihnen schaffen.
Wir laden zum Trainingsworkshop Datenanalyse ein. Angetrieben durch die Digitalisierung und die sogenannte Industrie 4.0 steigt die Erwartungshaltung gegenüber der Nutzung von Daten. Die Unsicherheit, Widersprüchlichkeit und Redundanz separat verarbeiteter einzelner Quellen soll durch die synergetische Zusammenführung heterogener Datensätze überwunden werden. Dabei steht der zunehmenden Aufgabenkomplexität eine ebenso zunehmende Verfügbarkeit an Daten und Analyseverfahren gegenüber.
Der Workshop vermittelt ein konzeptionelles Verständnis für moderne Datenanalyseverfahren (Machine Learning (ML), Multivariate Statistik) und soll durch ein anschließendes Hands-On Training mit Python (https://www.python.org) einen einfachen Einstieg in die Thematik ermöglichen. Der Kurs richtet sich an den wissenschaftlichen Nachwuchs.
Machine learning based multi-sensor fusion for the nondestructive testing of corrosion in concrete
(2020)
Half-cell potential mapping (HP) is the most popular non-destructive testing method (NDT) for locating corrosion damage in concrete. It is generally accepted that HP is susceptible to environmental factors caused by salt-related deterioration, such as different moisture and chloride gradients. Additional NDT methods are able to identify distinctive areas but are not yet used to estimate more accurate test results. We present a Supervised Machine Learning (SML) based approach to data fusion of seven different signal features to obtain higher quality information. SMLs are methods that explore (or learn) relationships between different (sensor) data from predefined data labels. To obtain a representative, labelled data set we conducted a comprehensive experiment simulating the deterioration cycle of a chloride exposed device in the laboratory. Our data set consists of 18 measurement campaigns, each containing HP, Ground Penetrating- Radar, Microwave Moisture and Wenner resistivity data. We compare the performance of different ML approaches. Many outperform the best single method, HP. We describe the intrinsic challenges posed by a data-driven approach in NDT and show how future work can help overcome them.
Visualization of automated multi‐sensor NDT assessment of concrete structures (NDT Data Fusion)
(2017)
Nondestructive testing(NDT) of concrete buildings allows efficient repair measures. Multi-sensor platforms collect large data sets but the data analysis is typically performed manually. Data Fusion uses the full potential of a multi-sensory data set in order to:
- improve information quality (reliability, robustness, accuracy, clarity, completeness)
- enable automated algorithm based data analysis.
This poster summarizes results of the IGSTC-project entitled "NDT-Data Fusion" (short title) and demonstrates a significant improvement in the testing performance for the example of corrosion detection.
Angetrieben durch Digitalisierung und die sogenannte Industrie 4.0 steigt die Erwartungshaltung gegenüber der Nutzung von Sensorik. Die Unsicherheit, Widersprüchlichkeit und Redundanz separat verarbeiteter einzelner Quellen soll durch die synergetische Zusammenführung heterogener Datensätze überwunden werden. Dabei steht der zunehmenden Aufgabenkomplexität eine ebenso zunehmende Verfügbarkeit an Sensoren, Daten und Rechenleistung gegenüber.
Der Begriff Datenfusion fasst Ansätze zusammen, die Daten zu abstrakteren, aber besser verständlichen Informationen verarbeiten. Die Fusion ist Kernbestandteil effektiver Assistenzsysteme und beweist in vielzähligen Aufgaben - von militärischen Anwendungen über Flug- und Fahrassistenzsysteme, bis in den Heimbereich – ihr großes Potenzial. Durch die wachsende Automatisierung bei der Messdatenerfassung wird Datenfusion auch in der industriellen Qualitätsprüfung und –sicherung zunehmend attraktiver.
Der Vortrag gibt einen Überblick über den breiten Themenkomplex und widmet sich dabei speziell der Fragen, welche Informationen in multivariaten Datensätzen stecken und wie sie extrahiert werden können. Abschließend wird ein Beispiel für die erfolgreiche Anwendung zur zerstörungsfreien Prüfung von Betonbauteilen vorgestellt. Der systematische Vergleich von Algorithmen, die Charakteristika des Labordatensatzes in unterschiedlicher Weise adressieren erlaubt überraschende Schlussfolgerungen.
Half-Cell-Potential Mapping (HP) is the most popular non-destructive testing (NDT) method for the detection of active corrosion in reinforced concrete. HP is influenced by parameters such as moisture and chloride gradients in the component. The sensitivity to the spatially small, but dangerous pitting is low. In this study we show how additional measurement information can be used with multi-sensor data fusion to improve the detection performance and to automate data evaluation. The fusion is based on supervised machine learning (SML). SML are methods that recognize relationships in (sensor) data based on given labels. We use SML to distinguish "defective" and "intact" labeled areas in our dataset. It consists of 18 measurement - each contains HP, ground radar, microwave moisture and Wenner resistivity data. Exact labels for changing environmental conditions were determined in a laboratory study on a reinforced concrete slab, which deteriorated controlled and accelerated. The deterioration progress was monitored continuously and corrosion was generated targeted at a predefined location. The detection results are quantified and statistically evaluated. The data fusion shows a significant improvement over the best single method (HP). We describe the challenges of data-driven approaches in nondestructive testing and show possible solutions.
This paper presents a novel approach for developing sustainable building materials through Sequential Learning. Data sets with a total of 1367 formulations of different types of alkali-activated building materials, including fly ash and blast furnace slag-based concrete and their respective compressive strength and CO2-footprint, were compiled from the literature to develop and evaluate this approach. Utilizing this data, a comprehensive computational study was undertaken to evaluate the efficacy of the proposed material design methodologies, simulating laboratory conditions reflective of real-world scenarios. The results indicate a significant reduction in development time and lower research costs enabled through predictions with machine learning. This work challenges common practices in data-driven materials development for building materials. Our results show, training data required for data-driven design may be much less than commonly suggested. Further, it is more important to establish a practical design framework than to choose more accurate models. This approach can be immediately implemented into practical applications and can be translated into significant advances in sustainable building materials development.
The amount of data generated worldwide is constantly increasing. These data come from a wide variety of sources and systems, are processed differently, have a multitude of formats, and are stored in an untraceable and unstructured manner, predominantly in natural language in data silos. This problem can be equally applied to the heterogeneous research data from materials science and engineering. In this domain, ways and solutions are increasingly being generated to smartly link material data together with their contextual information in a uniform and well-structured manner on platforms, thus making them discoverable, retrievable, and reusable for research and industry. Ontologies play a key role in this context. They enable the sustainable representation of expert knowledge and the semantically structured filling of databases with computer-processable data triples.
In this perspective article, we present the project initiative Materials-open-Laboratory (Mat-o-Lab) that aims to provide a collaborative environment for domain experts to digitize their research results and processes and make them fit for data-driven materials research and development. The overarching challenge is to generate connection points to further link data from other domains to harness the promised potential of big materials data and harvest new knowledge.
This data article introduces a dataset comprising 1630 alkali-activated concrete (AAC) mixes, compiled from 106 literature sources. The dataset underwent extensive curation to address feature redundancy, transcription errors, and duplicate data, yielding refined data ready for further data-driven science in the field of AAC, where this effort constitutes a novelty. The carbon footprint associated with each material used in the AAC mixes, as well as the corresponding CO2 footprint of every mix, were approximated using two published articles. Serving as a foundation for future expansions and rigorous data applications, this dataset enables the characterization of AAC properties through machine learning algorithms or as a benchmark for performance comparison among different formulations. In summary, the dataset provides a resource for researchers focusing on AAC and related materials and offers insights into the environmental benefits of substituting traditional Portland concrete with AAC.
In recent decades, the number of components in concrete has grown, particularly in formulations aimed at reducing carbon footprints. Innovations include diverse binders, supplementary cementitious materials, activators, concrete admixtures, and recycled aggregates. These developments target not only the enhancement of material properties but also the mitigation of the ecological and economic impacts of concrete — the most extensively used material by humankind. However, these advancements also introduce a greater variability in the composition of raw materials. The material’s behavior is significantly influenced by its nanoscale properties, which can pose challenges in accurate characterization. Consequently, there’s an increasing need for experimental tuning of formulations. This is accompanied by a more inconsistent composition of raw materials, which makes an experimental tuning of formulations more and more necessary. However, the increased complexity in composition presents a challenge in finding the ideal formulation through trial and error. Inverse design (ID) techniques offer a solution to this challenge by allowing for a comprehensive search of the entire design space to create new and improved concrete formulations. In this publication, we introduce the concept of ID and demonstrate how our open-source app “SLAMD” provides all necessary steps of the workflow to adapt it in the laboratory, lowering the application barriers. The intelligent screening process, guided by a predictive model, leads to a more efficient and effective data-driven material design process resulting in reduced carbon footprint and improved material quality while considering socio-economic factors in the materials design.
AbstractHigh-strength aluminum alloys used in aerospace and automotive applications obtain their strength through precipitation hardening. Achieving the desired mechanical properties requires precise control over the nanometer-sized precipitates. However, the microstructure of these alloys changes over time due to aging, leading to a deterioration in strength. Typically, the size, number, and distribution of precipitates for a quantitative assessment of microstructural changes are determined by manual analysis, which is subjective and time-consuming. In our work, we introduce a progressive and automatable approach that enables a more efficient, objective, and reproducible analysis of precipitates. The method involves several sequential steps using an image repository containing dark-field transmission electron microscopy (DF-TEM) images depicting various aging states of an aluminum alloy. During the process, precipitation contours are generated and quantitatively evaluated, and the results are comprehensibly transferred into semantic data structures. The use and deployment of Jupyter Notebooks, along with the beneficial implementation of Semantic Web technologies, significantly enhances the reproducibility and comparability of the findings. This work serves as an exemplar of FAIR image and research data management.
Large-language models (LLMs) such as GPT-4 caught the interest of many scientists. Recent studies suggested that these models could be useful in chemistry and materials science. To explore these possibilities, we organized a hackathon. This article chronicles the projects built as part of this hackathon. Participants employed LLMs for various applications, including predicting properties of molecules and materials, designing novel interfaces for tools, extracting knowledge from unstructured data, and developing new educational applications. The diverse topics and the fact that working prototypes could be generated in less than two days highlight that LLMs will profoundly impact the future of our fields. The rich collection of ideas and projects also indicates that the applications of LLMs are not limited to materials science and chemistry but offer potential benefits to a wide range of scientific disciplines.