| Name | opf_dataset_utils JSON |
| Version |
0.0.2
JSON |
| download |
| home_page | None |
| Summary | Physical calculations, plotting, and other utils for the PyG OPFDataset. |
| upload_time | 2024-10-19 12:20:38 |
| maintainer | None |
| docs_url | None |
| author | Viktor Todosijevic |
| requires_python | <3.12,>=3.10 |
| license | None |
| keywords |
|
| VCS |
|
| bugtrack_url |
|
| requirements |
No requirements were recorded.
|
| Travis-CI |
No Travis.
|
| coveralls test coverage |
No coveralls.
|
# opf-dataset-utils
In this package we provide utils to support working with the
[OPFDataset](https://pytorch-geometric.readthedocs.io/en/latest/generated/torch_geometric.datasets.OPFDataset.html#torch_geometric.datasets.OPFDataset).
We implement:
* Efficient OPF related calculations to be used as metrics or in a physics informed setting:
* Power Flow errors
* Branch powers
* Costs
* Inequalities
* Data visualization
* Enums for indexing the OPFData JSON format
* And more...
## Installation
### Requirements
Requires [_PyTorch_](https://pytorch.org/get-started/locally/) (any [version](https://pytorch.org/get-started/previous-versions/)
supported by PyTorch Geometric) and [_PyTorch Geometric_](https://pytorch-geometric.readthedocs.io/en/latest/install/installation.html) (>=2.6.0)
including it's optional dependencies (`pyg_lib`, `torch_scatter`, `torch_sparse`, `torch_cluster`, and `torch_spline_conv`).
May require installing _graphviz_ for visualization purposes.
### From PyPI
Install latest release from PyPI:
```
pip install opf_dataset_utils
```
### Directly from GitHub
Install the latest version directly from GitHub:
```
pip install git+https://github.com/viktor-ktorvi/opf-dataset-utils.git
```
## Usage
### Plotting
See [scripts/draw.py](scripts/draw.py) for a full example.
```python
from opf_dataset_utils.plotting.draw import draw_graph
draw_graph(dataset[0], ax=ax, node_size=300)
```
<p align="center">
<img src="img/draw_example.png" alt="Example graph" width="600"/>
</p>
### OPF calculations
#### Power flow errors
See [scripts/power_flow_errors.py](scripts/power_flow_errors.py) for a full example.
```python
from opf_dataset_utils.physics.errors.power_flow import calculate_power_flow_errors
with torch.no_grad():
untrained_predictions = untrained_model(batch.x_dict, batch.edge_index_dict)
mean_abs_errors_solution = calculate_power_flow_errors(batch, batch.y_dict).abs().mean()
mean_abs_errors_untrained = calculate_power_flow_errors(batch, untrained_predictions).abs().mean()
```
Example results:
```
Mean power flow errors:
Solution: 1.28563e-06 [p.u.]
Untrained model prediction: 413350.84375 [p.u.]
```
#### Costs
See [scripts/costs.py](scripts/costs.py) for a full example.
```python
from opf_dataset_utils.costs import (
calculate_costs_per_generator,
calculate_costs_per_grid,
)
costs_per_grid = calculate_costs_per_grid(data, data.y_dict)
costs_per_generator = calculate_costs_per_generator(data, data.y_dict)
```
Example results:
```
Costs per grid [$/h]:
tensor([7564.9541, 8622.5996, 7247.7939, 7504.3018, 8446.8887, 7478.8228,
8023.4907], device='cuda:0')
Costs per generator [$/h]:
tensor([4003.2192, 3561.7349, 0.0000, 0.0000, 0.0000, 0.0000,
4035.1089, 4587.4902, 0.0000, 0.0000, 0.0000, 0.0000,
3992.6309, 3255.1628, 0.0000, 0.0000, 0.0000, 0.0000,
4005.5452, 3498.7563, 0.0000, 0.0000, 0.0000, 0.0000,
4045.0081, 4401.8809, 0.0000, 0.0000, 0.0000, 0.0000,
3994.2031, 3484.6196, 0.0000, 0.0000, 0.0000, 0.0000,
4021.5444, 4001.9463, 0.0000, 0.0000, 0.0000, 0.0000],
device='cuda:0')
```
#### Inequality violations
See [scripts/inequality_errors.py](scripts/inequality_errors.py) for a full example.
```python
from opf_dataset_utils.enumerations import EdgeTypes
from opf_dataset_utils.physics.errors.inequality.voltage import calculate_upper_voltage_angle_difference_errors
from opf_dataset_utils.physics.errors.inequality.generator_power import calculate_lower_active_power_errors
upper_voltage_angle_violations_transformer = calculate_upper_voltage_angle_difference_errors(data, data.y_dict, EdgeTypes.TRANSFORMER)
lower_active_power_generation_violations = calculate_lower_active_power_errors(data, data.y_dict)
# etc.
```
Example results:
```
Worst case violations:
Solution:
Upper Vm: 1.1920928955078125e-07 [p.u.]
Lower Vm: 0.0 [p.u.]
Upper Va diff. (transformers): 0.0 [rad]
Upper Va diff. (AC lines): 0.0 [rad]
Lower Va diff. (transformers): 0.0 [rad]
Lower Va diff. (AC lines): 0.0 [rad]
Upper Pg: 0.0 [p.u.]
Lower Pg: 9.985742011053844e-09 [p.u.]
Upper Qg: 0.0 [p.u.]
Lower Qg: 0.0 [p.u.]
Upper S_ij (transformers): 0.0 [p.u.]
Upper S_ij (AC lines): 0.0 [p.u.]
Upper S_ji (transformers): 0.0 [p.u.]
Upper S_ji (AC lines): 0.0 [p.u.]
Untrained model:
Upper Vm: 112.72354125976562 [p.u.]
Lower Vm: 306.87078857421875 [p.u.]
Upper Va diff. (transformers): 0.0 [rad]
Upper Va diff. (AC lines): 1149.0396728515625 [rad]
Lower Va diff. (transformers): 124.8240737915039 [rad]
Lower Va diff. (AC lines): 686.0045776367188 [rad]
Upper Pg: 0.0 [p.u.]
Lower Pg: 33.55585479736328 [p.u.]
Upper Qg: 0.1578141152858734 [p.u.]
Lower Qg: 157.2376708984375 [p.u.]
Upper S_ij (transformers): 417092.96875 [p.u.]
Upper S_ij (AC lines): 2345705.25 [p.u.]
Upper S_ji (transformers): 10831.6123046875 [p.u.]
Upper S_ji (AC lines): 3082554.75 [p.u.]
Process finished with exit code 0
```
#### Branch power flows
See [scripts/branch_powers.py](scripts/branch_powers.py) for a full example.
```python
from opf_dataset_utils.enumerations import EdgeTypes
from opf_dataset_utils.physics.power import calculate_branch_powers
ac_line_powers_from, ac_line_powers_to = calculate_branch_powers(batch, batch.y_dict, EdgeTypes.AC_LINE)
transformer_powers_from, transformer_powers_to = calculate_branch_powers(batch, batch.y_dict, EdgeTypes.TRANSFORMER)
```
Example results:
```
AC line power flows [p.u.]:
tensor([ 1.9860e+00-0.0678j, 8.5920e-01+0.0821j, 7.7557e-01-0.0108j,
5.6563e-01-0.0239j, 4.0300e-01+0.0017j, -2.6185e-01+0.1008j,
-6.7861e-01+0.1603j, 8.0781e-02+0.0468j, 8.7156e-02+0.0244j,
1.9349e-01+0.0730j, 2.4476e-07-0.1052j, 2.9223e-01+0.0275j,
5.5007e-02+0.0391j, 9.9253e-02+0.0241j, -4.9108e-02-0.0280j,
1.5681e-02+0.0090j, 6.4659e-02+0.0209j], device='cuda:0')
Transformer line power flows [p.u.]:
tensor([0.2922-0.0601j, 0.1681+0.0037j, 0.4683+0.1187j], device='cuda:0')
```
#### Etc.
Raw data
{
"_id": null,
"home_page": null,
"name": "opf_dataset_utils",
"maintainer": null,
"docs_url": null,
"requires_python": "<3.12,>=3.10",
"maintainer_email": null,
"keywords": null,
"author": "Viktor Todosijevic",
"author_email": "todosijevicviktor998@gmail.com",
"download_url": "https://files.pythonhosted.org/packages/55/99/5197b58f608dd702e5fb84b73b81ceb8b84708b73dd0f5fe9ddb081e33bf/opf_dataset_utils-0.0.2.tar.gz",
"platform": null,
"description": "# opf-dataset-utils\n\nIn this package we provide utils to support working with the\n[OPFDataset](https://pytorch-geometric.readthedocs.io/en/latest/generated/torch_geometric.datasets.OPFDataset.html#torch_geometric.datasets.OPFDataset).\n\nWe implement:\n\n* Efficient OPF related calculations to be used as metrics or in a physics informed setting:\n * Power Flow errors\n * Branch powers\n * Costs\n * Inequalities\n* Data visualization\n* Enums for indexing the OPFData JSON format\n* And more...\n\n## Installation\n\n### Requirements\n\nRequires [_PyTorch_](https://pytorch.org/get-started/locally/) (any [version](https://pytorch.org/get-started/previous-versions/)\nsupported by PyTorch Geometric) and [_PyTorch Geometric_](https://pytorch-geometric.readthedocs.io/en/latest/install/installation.html) (>=2.6.0)\nincluding it's optional dependencies (`pyg_lib`, `torch_scatter`, `torch_sparse`, `torch_cluster`, and `torch_spline_conv`).\n\nMay require installing _graphviz_ for visualization purposes.\n\n### From PyPI\n\nInstall latest release from PyPI:\n```\npip install opf_dataset_utils\n```\n\n### Directly from GitHub\n\nInstall the latest version directly from GitHub:\n```\npip install git+https://github.com/viktor-ktorvi/opf-dataset-utils.git\n```\n\n## Usage\n\n### Plotting\n\nSee [scripts/draw.py](scripts/draw.py) for a full example.\n\n```python\nfrom opf_dataset_utils.plotting.draw import draw_graph\n\ndraw_graph(dataset[0], ax=ax, node_size=300)\n```\n\n<p align=\"center\">\n<img src=\"img/draw_example.png\" alt=\"Example graph\" width=\"600\"/>\n</p>\n\n### OPF calculations\n\n#### Power flow errors\n\nSee [scripts/power_flow_errors.py](scripts/power_flow_errors.py) for a full example.\n\n```python\nfrom opf_dataset_utils.physics.errors.power_flow import calculate_power_flow_errors\n\nwith torch.no_grad():\n untrained_predictions = untrained_model(batch.x_dict, batch.edge_index_dict)\n\nmean_abs_errors_solution = calculate_power_flow_errors(batch, batch.y_dict).abs().mean()\nmean_abs_errors_untrained = calculate_power_flow_errors(batch, untrained_predictions).abs().mean()\n```\nExample results:\n```\nMean power flow errors:\n\tSolution: 1.28563e-06 [p.u.]\n\tUntrained model prediction: 413350.84375 [p.u.]\n```\n\n#### Costs\n\nSee [scripts/costs.py](scripts/costs.py) for a full example.\n\n```python\nfrom opf_dataset_utils.costs import (\n calculate_costs_per_generator,\n calculate_costs_per_grid,\n)\n\ncosts_per_grid = calculate_costs_per_grid(data, data.y_dict)\ncosts_per_generator = calculate_costs_per_generator(data, data.y_dict)\n```\n\nExample results:\n\n```\nCosts per grid [$/h]:\ntensor([7564.9541, 8622.5996, 7247.7939, 7504.3018, 8446.8887, 7478.8228,\n 8023.4907], device='cuda:0')\nCosts per generator [$/h]:\ntensor([4003.2192, 3561.7349, 0.0000, 0.0000, 0.0000, 0.0000,\n 4035.1089, 4587.4902, 0.0000, 0.0000, 0.0000, 0.0000,\n 3992.6309, 3255.1628, 0.0000, 0.0000, 0.0000, 0.0000,\n 4005.5452, 3498.7563, 0.0000, 0.0000, 0.0000, 0.0000,\n 4045.0081, 4401.8809, 0.0000, 0.0000, 0.0000, 0.0000,\n 3994.2031, 3484.6196, 0.0000, 0.0000, 0.0000, 0.0000,\n 4021.5444, 4001.9463, 0.0000, 0.0000, 0.0000, 0.0000],\n device='cuda:0')\n```\n\n#### Inequality violations\n\nSee [scripts/inequality_errors.py](scripts/inequality_errors.py) for a full example.\n\n```python\nfrom opf_dataset_utils.enumerations import EdgeTypes\nfrom opf_dataset_utils.physics.errors.inequality.voltage import calculate_upper_voltage_angle_difference_errors\nfrom opf_dataset_utils.physics.errors.inequality.generator_power import calculate_lower_active_power_errors\n\nupper_voltage_angle_violations_transformer = calculate_upper_voltage_angle_difference_errors(data, data.y_dict, EdgeTypes.TRANSFORMER)\nlower_active_power_generation_violations = calculate_lower_active_power_errors(data, data.y_dict)\n# etc.\n```\n\nExample results:\n\n```\nWorst case violations:\n\nSolution:\n\nUpper Vm: 1.1920928955078125e-07 [p.u.] \nLower Vm: 0.0 [p.u.] \nUpper Va diff. (transformers): 0.0 [rad] \nUpper Va diff. (AC lines): 0.0 [rad] \nLower Va diff. (transformers): 0.0 [rad] \nLower Va diff. (AC lines): 0.0 [rad] \nUpper Pg: 0.0 [p.u.] \nLower Pg: 9.985742011053844e-09 [p.u.] \nUpper Qg: 0.0 [p.u.] \nLower Qg: 0.0 [p.u.] \nUpper S_ij (transformers): 0.0 [p.u.] \nUpper S_ij (AC lines): 0.0 [p.u.] \nUpper S_ji (transformers): 0.0 [p.u.] \nUpper S_ji (AC lines): 0.0 [p.u.] \n\nUntrained model:\n\nUpper Vm: 112.72354125976562 [p.u.] \nLower Vm: 306.87078857421875 [p.u.] \nUpper Va diff. (transformers): 0.0 [rad] \nUpper Va diff. (AC lines): 1149.0396728515625 [rad] \nLower Va diff. (transformers): 124.8240737915039 [rad] \nLower Va diff. (AC lines): 686.0045776367188 [rad] \nUpper Pg: 0.0 [p.u.] \nLower Pg: 33.55585479736328 [p.u.] \nUpper Qg: 0.1578141152858734 [p.u.] \nLower Qg: 157.2376708984375 [p.u.] \nUpper S_ij (transformers): 417092.96875 [p.u.] \nUpper S_ij (AC lines): 2345705.25 [p.u.] \nUpper S_ji (transformers): 10831.6123046875 [p.u.] \nUpper S_ji (AC lines): 3082554.75 [p.u.] \n\nProcess finished with exit code 0\n\n```\n\n#### Branch power flows\n\nSee [scripts/branch_powers.py](scripts/branch_powers.py) for a full example.\n\n```python\nfrom opf_dataset_utils.enumerations import EdgeTypes\nfrom opf_dataset_utils.physics.power import calculate_branch_powers\n\nac_line_powers_from, ac_line_powers_to = calculate_branch_powers(batch, batch.y_dict, EdgeTypes.AC_LINE)\ntransformer_powers_from, transformer_powers_to = calculate_branch_powers(batch, batch.y_dict, EdgeTypes.TRANSFORMER)\n```\n\nExample results:\n\n```\nAC line power flows [p.u.]:\ntensor([ 1.9860e+00-0.0678j, 8.5920e-01+0.0821j, 7.7557e-01-0.0108j,\n 5.6563e-01-0.0239j, 4.0300e-01+0.0017j, -2.6185e-01+0.1008j,\n -6.7861e-01+0.1603j, 8.0781e-02+0.0468j, 8.7156e-02+0.0244j,\n 1.9349e-01+0.0730j, 2.4476e-07-0.1052j, 2.9223e-01+0.0275j,\n 5.5007e-02+0.0391j, 9.9253e-02+0.0241j, -4.9108e-02-0.0280j,\n 1.5681e-02+0.0090j, 6.4659e-02+0.0209j], device='cuda:0')\n\n\nTransformer line power flows [p.u.]:\ntensor([0.2922-0.0601j, 0.1681+0.0037j, 0.4683+0.1187j], device='cuda:0')\n```\n\n#### Etc.\n",
"bugtrack_url": null,
"license": null,
"summary": "Physical calculations, plotting, and other utils for the PyG OPFDataset.",
"version": "0.0.2",
"project_urls": null,
"split_keywords": [],
"urls": [
{
"comment_text": "",
"digests": {
"blake2b_256": "c007bb1272009bb3d9fb01150deb28f55f4e5eb069e57091fdb8b8a9b7a7015f",
"md5": "48d7a168fc4e67248a91f2ccefcbdcc9",
"sha256": "c64c36d35e7da78e6f9244d6dbc6c361f86c8534f9ff49a5a481ce96ad71916a"
},
"downloads": -1,
"filename": "opf_dataset_utils-0.0.2-py3-none-any.whl",
"has_sig": false,
"md5_digest": "48d7a168fc4e67248a91f2ccefcbdcc9",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": "<3.12,>=3.10",
"size": 18991,
"upload_time": "2024-10-19T12:20:36",
"upload_time_iso_8601": "2024-10-19T12:20:36.586769Z",
"url": "https://files.pythonhosted.org/packages/c0/07/bb1272009bb3d9fb01150deb28f55f4e5eb069e57091fdb8b8a9b7a7015f/opf_dataset_utils-0.0.2-py3-none-any.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": "",
"digests": {
"blake2b_256": "55995197b58f608dd702e5fb84b73b81ceb8b84708b73dd0f5fe9ddb081e33bf",
"md5": "ca937c3e883b2bd8e1df937f46c2f053",
"sha256": "beb38818bacc6938d758c5c4492c695020185abcc8dfad36c87c6a703332a2d0"
},
"downloads": -1,
"filename": "opf_dataset_utils-0.0.2.tar.gz",
"has_sig": false,
"md5_digest": "ca937c3e883b2bd8e1df937f46c2f053",
"packagetype": "sdist",
"python_version": "source",
"requires_python": "<3.12,>=3.10",
"size": 30781,
"upload_time": "2024-10-19T12:20:38",
"upload_time_iso_8601": "2024-10-19T12:20:38.346194Z",
"url": "https://files.pythonhosted.org/packages/55/99/5197b58f608dd702e5fb84b73b81ceb8b84708b73dd0f5fe9ddb081e33bf/opf_dataset_utils-0.0.2.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2024-10-19 12:20:38",
"github": false,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"lcname": "opf_dataset_utils"
}
