TBM_advance_classification

Code and data repository for the paper Challenges and Opportunities of Data Driven Advance Classification of Hard Rock TBMs by Georg H. Erharter¹, Paul Unterlaß², Nedim Radončić³, Thomas Marcher², Jamal Rostami⁴

1. Norwegian Geotechnical Institute, Sandakerveien 140, Oslo, Norway
2. Institute of Rock Mechanics and Tunnelling, Graz University of Technology, Rechbauerstraße 12, Graz, Austria
3. iC Consulenten ZT GmbH, Schönbrunnerstraße 12, Vienna, Austria
4. Colorado School of Mines, 1500 Illinois St, Golden, Colorado, United States of America

• correspondence: georg.erharter@ngi.no

The paper is currently in the review phase.

Code authors: Georg H. Erharter, Paul Unterlaß & Theresa Maier

Benchmark

The advance classification of the 3 generated synthetic TBM operational data sets can be performed using the DATA_02_analyzer.py file in the src folder. To analyze the data set of a specific TBM (A, B, C), the letter X in the code 'SAMPLE = 'TBM_X' in the section '# fixed values and variables' must be replaced by the letter of the TBM to be analyzed. As a result, an .xlsx file (TBM_X_2_synthetic_strokes) is saved in the folder data. An advance class of 0 indicates regular advance and an advance class of 1 indicates exceptional advance. In addition, a status message is displayed giving the arithmetic mean and the median of the stroke for regular and exceptional advance.

Below, the expected strokes for regular and exceptional advance are listed for each of the 3 synthetic data sets (TBM A, -B, -C).

Expected result:

TBM A:

• Regular advance:
  • n strokes based on arithmetic mean: 473
  • n strokes based on median: 464
• Exceptional advance:
  • n strokes based on arithmetic mean: 116
  • n strokes based on median: 125

TBM B:

• Regular advance:
  • n strokes based on arithmetic mean: 546
  • n strokes based on median: 556
• Exceptional advance:
  • n strokes based on arithmetic mean: 43
  • n strokes based on median: 33

TBM C:

• Regular advance:
  • n strokes based on arithmetic mean: 491
  • n strokes based on median: 469
• Exceptional advance:
  • n strokes based on arithmetic mean: 98
  • n strokes based on median: 120

Synthetic TBM operational data

The synthetic Tunnel Boring Machine (TBM) operational data can be found in the folder "data". Datasets for 3 different TBMs were generated which are denoted TBM A, -B, -C. The data was synthezised using generative adverserial networks (GANs) based on real TBM operational data.

For each TBM 4 different files are given (replace X with A, B, C for the different TBMs):

• TBM_X_0_synthetic_raw: direct output of the GANs in vectors of length 4096. DO NOT USE FOR GEOTECHNICAL PURPOSES. Type: parquet file. See section 2.2.1 in paper.
• TBM_X_1_synthetic_realistic.zip: Synthetic TBM data after post processing. Type: zipped .csv file. See section 2.2.1 in paper.
• TBM_X_2_synthetic_advance.xlsx: Synthetic TBM operational data after basic data cleaning where standstills have been removed. Type: excel file. See section 3.1 in paper.
• TBM_X_2_synthetic_strokes.xlsx: Finally processed TBM operational data with stroke-wise aggregation. Type: excel file. See section 3.5 in paper.

Requirements

The environment is set up using conda.

To do this create an environment called TBM_data 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 Jv

contact

georg.erharter@ngi.no