Update README.md
Browse files
README.md
CHANGED
|
@@ -40,7 +40,6 @@ MDOF/
|
|
| 40 |
└── Data/
|
| 41 |
└── fno/
|
| 42 |
├── Blg_F6_18m_IM7_st0.h5.h5 # Floor acc. response
|
| 43 |
-
└── ...
|
| 44 |
```
|
| 45 |
|
| 46 |
### Ground Motion File (`GMs_knet_3474_AF_57.h5`)
|
|
@@ -65,69 +64,7 @@ Each file stores the simulated structural response for all GM × scale combinati
|
|
| 65 |
|
| 66 |
## Loading
|
| 67 |
|
| 68 |
-
|
| 69 |
-
|
| 70 |
-
```python
|
| 71 |
-
import h5py
|
| 72 |
-
import numpy as np
|
| 73 |
-
|
| 74 |
-
GM_FILE = "path/to/MDOF/All_GMs/GMs_knet_3474_AF_57.h5"
|
| 75 |
-
|
| 76 |
-
with h5py.File(GM_FILE, "r") as f:
|
| 77 |
-
# Load ground motion i=0, amplitude factor j=0
|
| 78 |
-
gm = f["gm_0/af_0/data"][:] # shape (3000,)
|
| 79 |
-
pga = f["gm_0/af_0/pga"][()]
|
| 80 |
-
|
| 81 |
-
print(f"GM shape: {gm.shape}, PGA: {pga:.4f} m/s²")
|
| 82 |
-
```
|
| 83 |
-
|
| 84 |
-
```python
|
| 85 |
-
import h5py
|
| 86 |
-
import numpy as np
|
| 87 |
-
|
| 88 |
-
BLG_FILE = "path/to/MDOF/knet-250/Data/fno/building_0001.h5"
|
| 89 |
-
|
| 90 |
-
with h5py.File(BLG_FILE, "r") as f:
|
| 91 |
-
# Floor acceleration response for GM i=0, AF j=0
|
| 92 |
-
floor_acc = f["response/gm_0/af_0/floor_acc"][:] # (n_floors, 3000)
|
| 93 |
-
damage_state = f["building/damage_state/gm_0/af_0"][()]
|
| 94 |
-
|
| 95 |
-
print(f"Floor acc shape: {floor_acc.shape}, Damage state: {damage_state}")
|
| 96 |
-
```
|
| 97 |
-
|
| 98 |
-
### Option 2 — PyTorch Dataset (recommended for training)
|
| 99 |
-
|
| 100 |
-
Clone the [SeismicFNO](https://github.com/HKUJasonJiang/Seismic-FNO) repository and use `DynamicDataset`:
|
| 101 |
-
|
| 102 |
-
```python
|
| 103 |
-
import numpy as np
|
| 104 |
-
from torch.utils.data import DataLoader
|
| 105 |
-
from module.dataprep_v2 import DynamicDataset
|
| 106 |
-
|
| 107 |
-
GM_FILE = "path/to/MDOF/All_GMs/GMs_knet_3474_AF_57.h5"
|
| 108 |
-
BUILDING_DIR = "path/to/MDOF/knet-250/Data/fno/"
|
| 109 |
-
|
| 110 |
-
# Full dataset (all 3474 × 57 combinations)
|
| 111 |
-
dataset = DynamicDataset(
|
| 112 |
-
gm_file_path = GM_FILE,
|
| 113 |
-
building_files_dir = BUILDING_DIR,
|
| 114 |
-
)
|
| 115 |
-
|
| 116 |
-
# Or pass a pre-computed index array for train/val/test splits
|
| 117 |
-
rng = np.random.default_rng(42)
|
| 118 |
-
indices = rng.permutation(len(dataset))
|
| 119 |
-
train_ds = DynamicDataset(GM_FILE, BUILDING_DIR, gm_indices=indices[:int(0.7 * len(indices))])
|
| 120 |
-
|
| 121 |
-
loader = DataLoader(train_ds, batch_size=64, shuffle=True, num_workers=4)
|
| 122 |
-
|
| 123 |
-
# Each batch: (gm, building_attributes, floor_acc_response, damage_state)
|
| 124 |
-
gm, attr, resp, ds = next(iter(loader))
|
| 125 |
-
print(gm.shape, resp.shape) # (64, 3000, 1), (64, 3000, 1)
|
| 126 |
-
```
|
| 127 |
-
|
| 128 |
-
## Citation
|
| 129 |
-
|
| 130 |
-
If you use this dataset, please cite the K-NET strong-motion network and the associated SeismicFNO paper (forthcoming).
|
| 131 |
|
| 132 |
## License
|
| 133 |
|
|
|
|
| 40 |
└── Data/
|
| 41 |
└── fno/
|
| 42 |
├── Blg_F6_18m_IM7_st0.h5.h5 # Floor acc. response
|
|
|
|
| 43 |
```
|
| 44 |
|
| 45 |
### Ground Motion File (`GMs_knet_3474_AF_57.h5`)
|
|
|
|
| 64 |
|
| 65 |
## Loading
|
| 66 |
|
| 67 |
+
Refer to github repo and codes.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 68 |
|
| 69 |
## License
|
| 70 |
|