TubeBEND

https://github.com/zeyneddinoz/tubebend
Admin
TubeBEND is a benchmark dataset and Python toolkit containing real-world sensor, setup, and geometry...
Tokens:3,220
Snippets:25
Trust Score:5.6
Update:2 weeks ago
Show doc for...
Tokens:
Raw
### Install dependencies with pip

Source: https://github.com/zeyneddinoz/tubebend/blob/main/README.md

Install the required packages for the dataset. Python version 3.10 or higher is required.

```bash
pip install -r requirements.txt
```

--------------------------------

### Show all bending setups with take_all_bending_setups

Source: https://github.com/zeyneddinoz/tubebend/blob/main/README.md

Imports and calls take_all_bending_setups from the utils module to retrieve all bending setups from the loaded dictionary. The result is displayed as a table with columns for experiment, tube, outer-diameter, wall-thickness, target-angle, and various die/mandrel positions.

```python
# To see all bending setups in one table
from utils import take_all_bending_setups

all_bending_setups = take_all_bending_setups(loaded_dict)
all_bending_setups:
```

--------------------------------

### Load experiment data as dictionary with loaded_dict[f'Exp_{experiment_number}']

Source: https://github.com/zeyneddinoz/tubebend/blob/main/README.md

Loads a specific experiment's data as a dictionary from the loaded_dict using the key 'Exp_<number>'. Experiment numbers range from 1 to 318; the example selects experiment 11.

```python
# Experiment numbers range from 1 to 318. For example, we select the 11th one:
experiment_number = 11

experiment_as_dictinary = loaded_dict[f'Exp_{experiment_number}']
```

--------------------------------

### Load dataset with pickle

Source: https://github.com/zeyneddinoz/tubebend/blob/main/README.md

Load the serialized pickle file containing the dataset. The file is tracked with Git LFS and is about 800 MB.

```python
import pickle

with open('experiments_process_and_results.pkl', 'rb') as f:
    loaded_dict = pickle.load(f)
```

--------------------------------

### Load sensor loads section from experiment dataframe

Source: https://github.com/zeyneddinoz/tubebend/blob/main/README.md

Loads a specific section from the selected experiment's dataframe using the section name 'process_parameters_loads_sensor'. The resulting dataframe contains time and sensor load columns (e.g., SENSOR_MANDREL_AXIAL_Load_[kN]).

```python
# Loading the desired section from the dataframe
features_as_pandas_dataframe = experiment_as_dictinary['process_parameters_loads_sensor']
```

--------------------------------

### Cite TubeBEND dataset

Source: https://github.com/zeyneddinoz/tubebend/blob/main/README.md

BibTeX entry for citing the TubeBEND dataset paper.

```bibtex
@article{oz2025tubebend,
  title={TubeBEND: A Real-World Dataset for Geometry Prediction in Rotary Draw Bending},
  author={Oz, Zeyneddin and Knoche, Jonas and Yazdani, Alireza and Engel, Bernd and Van Laerhoven, Kristof},
  journal={arXiv preprint arXiv:2509.10272},
  year={2025}
}
```

--------------------------------

### Plot one sensor feature with multi_sensor_subplots

Source: https://github.com/zeyneddinoz/tubebend/blob/main/README.md

Plots a single sensor column (e.g., 'SENSOR_MANDREL_AXIAL_Load_[kN]') from the DataFrame by selecting it along with the time column. Saves the figure and prints the available column names for other sensor options.

```python
sensor_name = 'SENSOR_MANDREL_AXIAL_Load_[kN]'

multi_sensor_subplots(features_as_pandas_dataframe[['Time_[s]', sensor_name]], save_fig=True, output_path=f"Exp_{experiment_number}_{sensor_name}_sensors")

# Other options can be seen in this list:
print(list(features_as_pandas_dataframe.columns)[1:])
```

--------------------------------

### Plot all features with multi_sensor_subplots

Source: https://github.com/zeyneddinoz/tubebend/blob/main/README.md

Plots all features from a given pandas DataFrame using the multi_sensor_subplots utility. Saves the figure to the specified output path and returns an interactive plot (viewable in Firefox only).

```python
from utils import multi_sensor_subplots

multi_sensor_subplots(features_as_pandas_dataframe, save_fig=True, output_path=f"Exp_{experiment_number}_all_sensors")
```
Tokens:
Raw