TBM_anomaly_detection

Code and data repository for the paper TBM Operational Data-Driven Anomaly Detection in Hard Rock Excavations by Paul J. Unterlaß¹, Mario Wölflingseder¹, Thomas Marcher¹ submitted for publication at the 9th of April 2025 to the journal "Tunnelling and Underground Space Technology".

1. Institute of Rock Mechanics and Tunnelling, Graz University of Technology, Rechbauerstraße 12, Graz, Austria

• correspondence: unterlass@tugraz.at

Code authors: Paul J. Unterlass & Mario Wölflingseder

Synthetic TBM operational data

The synthetic TBM operational data can be found in the folder "data". Datasets for 2 different TBMs are available, denoted as TBM A, -B. The data was synthezised using generative adverserial networks (GANs) based on real TBM operational data. Further synthetic data and the code of GANs can be found in the following Github repository: https://github.com/geograz/TBM_advance_classification Further information on the synthetic data can be found in the following publications: https://doi.org/10.1007/s00603-025-04542-4 (open access) and https://doi.org/10.1007/978-3-031-20241-4_1

TBM anomaly detection code

In the folder src the code for pre-processing of the datasets, the variational autoencoder (VAE) model, the VAE training/validation/testing pipeline and the alternative anomaly detection techniques can be found.

Pre-trained VAE models and results

In the folder results various pre-trained VAE models and the anomaly detection results can be found.

Requirements

The environment is set up using conda.

To do this create an environment called TBM_anomaly_detection using environment.yaml with the help of conda. If you get pip errors, install pip libraries manually, e.g. pip install pandas

conda env create --file environment.yaml

Activate the new environment with:

conda activate TBM_anomaly_detection

contact

unterlass@tugraz.at