Short description

For infrastructure monitoring we use glas fiber cables as used for communications as an acoustic sensor [1,2]. This method provides us with »1000 virtual channels along up to 100~km. It can be viewed as a virtual microphone array along the fibres, determining the location of a specific microphone with the time-of-flight of a light pulse to the place and return.

Construction work in close vicinity of either pipelines or high-voltage power cables might damage the lines [3] and therefore cause high costs or even significant damage in case of gas explosions. In order to detect potentially dangerous events, we need to classify the data that is recorded. Most data will not be interesting, since dangerous events are avoided and therefore very rare. The aim of this project/thesis is to detect anomalous events in those audio channels close to real-time with a latency of around low one digit minutes and classify the type of event. Limited data of dangerous events exist so a model can be trained to determine the most likely cause of the anomaly.

[1] Lu, P., et al. (2019). Distributed optical fiber sensing: Review and perspective. Applied Physics Reviews, 6(4), 41302. https://doi.org/10.1063/1.5113955

[2] Ghazali, M. F., Mohamad, H., Nasir, M. Y. M., Hamzh, A., Abdullah, M. A., Aziz, N. F. A., Thansirichaisree, P., & Zan, M. S. D. (2024). State-of-The-Art application and challenges of optical fibre distributed acoustic sensing in civil engineering. Optical Fiber Technology, 87, 103911. https://doi.org/10.1016/j.yofte.2024.103911

[3] Zhu, H.-H., Liu, W., Wang, T., Su, J.-W., & Shi, B. (2022). Distributed Acoustic Sensing for Monitoring Linear Infrastructures: Current Status and Trends. Sensors, 22(19), 7550. https://doi.org/10.3390/s22197550

Your Tasks

• Literature research on audio classification with focus on DAS
• Implementation of classification algorithms
• Evaluation and Documentation

Your Profile/Prerequisites

• Motivation and interest in the topic
• Background in Signal Processing and Machine Learning
• Strong programming background, ideally in Python or Julia

Contact:

Martin Hagmüller (hagmueller@tugraz.at or 0316/873 4377)