Filtern
Dokumenttyp
- Zeitschriftenartikel (2)
- Beitrag zu einem Tagungsband (2)
- Vortrag (1)
- Forschungsdatensatz (1)
Schlagworte
- Radar (4)
- Feuchte (3)
- GPR (2)
- Machine Learning (2)
- Material Moisture (2)
- Neutronensonde (2)
- Building Floors (1)
- Building floor (1)
- Building floors (1)
- Civil Engineering (1)
Organisationseinheit der BAM
Eingeladener Vortrag
- nein (1)
Die Idee des Dissertationsprojekts ist der parallele Einsatz des Radarverfahrens und der Neutronensonde zur Lokalisierung von Feuchteschäden in Fußbodenaufbauten. Hierbei soll die integrale Messweise der Neutronensonde durch die vertikale Information des Radarsignals ergänzt werden, um zukünftig auf zerstörende Sondierungsbohrungen verzichten zu können. Primäres Ziel ist eine automatisierte und zerstörungsfreie Klassifizierung und Quantifizierung verschiedener Schadensfälle im Fußboden, welche zur Abschätzung und Auswahl des Sanierungsaufwands dienen soll. In systematischen Laborstudien an modular aufgebauten Referenzprobekörpern werden die Nachweisgrenzen der beiden Verfahren für unterschiedliche, häufig anzutreffende Fußbodenaufbauten untersucht. Hierfür wurde ein Satz verschiedenster Estrichprobekörper gefertigt und deren Hydratisierungsprozess gravimetrisch und mit den benannten Feuchtemessverfahren beobachtet. In der laufenden Auswertung konnten bereits signifikante Signalmerkmale der Radarmessungen extrahiert, sowie dessen Korrelation zum Feuchteverlauf der Proben gezeigt werden. Der modulare Aufbau der Fußbodenschichten zur Simulation von gängigen Feuchteschäden folgt im Anschluss. Über Datenfusion und Signalverarbeitung sollen so innovative Auswertungsansätze entwickelt und deren Validität an realen Schadensfällen geprüft werden.
Das Neutronensondenverfahren wird bereits seit vielen Jahren erfolgreich zur Eingrenzung und Quantifizierung auftretender Feuchteschäden an Fußböden eingesetzt. Hierzu bedarf es jedoch einer Vielzahl zerstörender Sondierungsbohrungen, welche die gewonnenen Messdaten kalibrieren und eine Tiefenzuordnung des Flüssigwassers zulassen. Dadurch entsteht ein zeitlicher und finanzieller Aufwand, der durch den parallelen Einsatz des elektromagnetischen Radarverfahrens vermieden werden könnte. Mit seiner hohen Sensitivität für Wasser bietet diese Messmethode die Möglichkeit der vertikalen Lokalisierung von Feuchte, was zu einer automatisierten Klassifizierung typischer Schadensfälle beitragen soll.
In einem laufenden Forschungsvorhaben werden in systematischen Laborstudien gängige Schadensfälle an häufig anzutreffenden Fußbodenaufbauten simuliert und deren Einfluss auf die genannten Verfahren untersucht. Hierbei kommen Zement- und Anhydritestriche, sowie unterschiedliche Dämmmaterialen mit variierenden Schichtdicken zum Einsatz. Wesentlicher Bestandteil der Auswertung ist die Extraktion signifikanter Signalmerkmale des Radarverfahrens, welche Rückschlüsse auf den Schadensfall und ggf. die Wassermenge zulassen. Weiterführend sollen die Kombinationsmöglichkeiten der verschiedenen Signalmerkmale und der Neutronensondendaten durch Methoden der multivariaten Datenauswertung und des maschinellen Lernens geprüft werden. Die Unabhängigkeit gegenüber wechselnden Schichtdicken und Materialien steht hierbei besonders im Fokus und soll anhand der erzielten Ergebnisse evaluiert werden.
In this ongoing research project, we study the influence of moisture damage on Ground Penetrating Radar (GPR) in different floor constructions. For this purpose, a measurement setup with interchangeable layers is developed to vary the screed material (cement or anhydrite) and insulation material (glass wool, perlite, expanded and extruded polystyrene), as well as the respective layer thickness. The evaluation of the 2 GHz common-offset radar measurements is focused on the extraction of distinctive signal features that can be used to classify the underlying case of damage without any further information about the hidden materials or layer thicknesses. In the collected dataset, we analyze the horizontal distribution of A-scan features in corresponding B-scans to detect water in the insulation layer. Furthermore, possible combinations of these features are investigated with the use of multivariate data analysis and machine learning (logistic regression) in order to evaluate the mutual dependencies. In this study, the combination of an amplitude- and frequency-based feature achieved an accuracy of 93.2 % and performed best to detect a damage in floor insulations.
To date, the destructive extraction and analysis of drilling cores is the main possibility to obtain depth information about damaging water ingress in building floors. The time- and costintensive procedure constitutes an additional burden for building insurances that already list piped water damage as their largest item. With its high sensitivity for water, a ground-penetrating radar (GPR) could provide important support to approach this problem in a non-destructive way. In this research, we study the influence of moisture damage on GPR signals at different floor constructions. For this purpose, a modular specimen with interchangeable layers is developed to vary the screed and insulation material, as well as the respective layer thickness. The obtained data set is then used to investigate suitable signal features to classify three scenarios: dry, damaged insulation, and damaged screed. It was found that analyzing statistical distributions of A-scan features inside one B-scan allows for accurate classification on unknown floor constructions. Combining the features with multivariate data analysis and machine learning was the key to achieve satisfying results. The developed method provides a basis for upcoming validations on real damage cases.
Machine learning in non-destructive testing (NDT) offers significant potential for efficient daily data analysis and uncovering previously unknown relationships in persistent problems. However, its successful application heavily depends on the availability of a diverse and well-labeled training dataset, which is often lacking, raising questions about the transferability of trained algorithms to new datasets. To examine this issue closely, the authors applied classifiers trained with laboratory Ground Penetrating Radar (GPR) data to categorize on-site moisture damage in layered building floors. The investigations were conducted at five different locations in Germany. For reference, cores were taken at each measurement point and labeled as (i) dry, (ii) with insulation damage, or (iii) with screed damage. Compared to the accuracies of 84 % to 90 % within the laboratory training data (504 B-Scans), the classifiers achieved a lower overall accuracy of 53 % for on-site data (72 B-Scans). This discrepancy is mainly attributable to a significantly higher dynamic of all signal features extracted from on-site measurements compared to laboratory training data. Nevertheless, this study highlights the promising sensitivity of GPR for identifying individual damage cases. In particular the results showing insulation damage, which cannot be detected by any other non-destructive method, revealed characteristic patterns. The accurate interpretation of such results still depends on trained personnel, whereby fully automated approaches would require a larger and diverse on-site data set. Until then, the findings of this work contribute to a more reliable analysis of moisture damage in building floors using GPR and offer practical insights into applying machine learning to non-destructive testing for civil engineering (NDT-CE).
Related work
Laboratory Study:
Combining Signal Features of Ground-Penetrating Radar to Classify Moisture Damage in Layered Building Floors
https://doi.org/10.3390/app11198820
On-Site Study:
TBA
Doctoral Thesis:
Non-destructive classification of moisture deterioration in layered building floors using ground penetrating radar
https://doi.org/10.14279/depositonce-19306
Measurement Parameters
The GPR measurements were carried out with the SIR 20 from GSSI and a 2 GHz antenna pair (bandwidth 1 GHz to 3 GHz) in common-offset configuration. Each B-Scan consists of N A-Scans, each including 512 samples of a 11 ns time window. Survey lines were recorded with 250 A-Scans/ meter, which equals a 4 mm spacing between each A-Scan No Gains were applied.
Folder Description:
Lab_dry, Lab_insulDamage, Lab_screedDamage
- each contain 168 Measurements (B-Scans) in .csv on 84 dry floors, floors with insulation damage and screed damage.
- each floor setup was measured twice on two orthogonal survey lines, indicated by _Line1_ and _Line2_ in the file name.
- the file names encode the building floor setup e.g. CT50XP100 describes a 50 mm cement screed with 100 mm extruded polystyrene below
- the material codes are
CT: cement screed, CA: anhydrite screed, EP: expanded polystyrene, XP: extruded polystyrene, GW: glass wool, PS: perlites
further information can be found in the publication https://doi.org/10.3390/app11198820
OnSite_
- 5 folders containing B-Scans on 5 different practical moisture damages
- the building floor setup is encoded according to the lab with an additional measurement point numbering at the start and a damage case annotation at the end of the file name with _dry, _insulationDamage and_screedDamage
File Description:
B-Scans, Measurement files - no header
- dimension: 512 x N data point with N beeing the number of A-Scans including 512 samples of a 11 ns time window.
- survey lines were recorded with 250 A-Scans/ meter, which equals a 4 mm spacing between each A-Scan
Moisture References
- Moist_Reference of On-Site Locations include the columns MeasPoint: Measurement point, wt%Screed: moisture content of screed layer in mass percent; wt%Insul: moisture content of insulation layer in mass percent. References were obtained by drilling cores with 68 mm diameter in the center of each survey line.
- Moist_Reference_Screed of Lab data include the columns Screed: Screed material and thickness in mm, wt%Screed moisture content of screed layer in mass percent
- Moist Reference_Insul of Lab data include the columns Insulation: Insulation material and thickness in mm, water addition in l: water added to the insulation layer in liters, V%Insulation: water added to the insulation layer in volume percent, RH%: resulting relative humidy in the insulation layer during measurement. These References are only avaible for Lab measurements on insulation damages.