University of Twente Student Theses
Explainable structural health monitoring(SHM) for damage classification based on vibration
Kongala, Sai Ganesh Reddy (2024) Explainable structural health monitoring(SHM) for damage classification based on vibration.
|
PDF
3MB |
| Abstract: | Traditional SHM methods often face challenges in scaling and accuracy due to their dependency on manual inspection and non-explainable ML models. Recent advancements in big data analytics, machine learning, cloud data storage systems, and wireless sensor network technologies have significantly expanded the potential for data-driven Structural health Monitoring (SHM) systems for large-scale engineering projects. The application of machine learning methods in SHM has seen considerable growth in recent years. However, given the cross-domain nature of this technology, there emerges a critical need for Explainable Artificial Intelligence (XAI) within the realm of SHM. This study investigates the application of XAI frameworks to enhance the interpretability and reliability of SHM systems, specifically targeting damage classification using vibration data. This research employs accelerometer sensor networks to collect vibration data from composite structures, utilizing various ML models such as Random Forest, Support Vector Machine, XGBoost, Convolutional Neural Networks (CNN), and Transformer networks. The study underscores the superior performance of advanced models, particularly CNNs and Transformer networks, in accurately identifying structural defects. The thesis also addresses the challenges of data corruption in SHM systems. Data corruption, caused by sensor faults, transmission errors, environmental interference, and software bugs, can lead to false positives and false negatives, severely impacting the reliability of SHM systems. To mitigate these effects, robust data validation and cleaning mechanisms are implemented, including anomaly detection algorithms. Future research directions include optimizing sensor coverage, integrating advanced sensing technologies, and fostering interdisciplinary collaborations to develop economically viable and technically proficient SHM solutions. |
| Item Type: | Essay (Master) |
| Faculty: | EEMCS: Electrical Engineering, Mathematics and Computer Science |
| Subject: | 54 computer science |
| Programme: | Business Information Technology MSc (60025) |
| Link to this item: | https://purl.utwente.nl/essays/99639 |
| Export this item as: | BibTeX EndNote HTML Citation Reference Manager |
Show download statistics for this publication
Show download statistics for this publicationRepository Staff Only: item control page