oma-python

Open-source Operational Modal Analysis (OMA) algorithms for Python. Estimate natural frequencies, damping ratios, and mode shapes from ambient vibration measurements.

Install the distribution as oma-python; import the library as dynoma.

Install

pip install oma-python

For local development with uv:

uv sync --group dev

Quick start (FDD)

Frequency Domain Decomposition on a synthetic multi-channel signal:

import numpy as np
from dynoma import FDD

rng = np.random.default_rng(0)
fs = 100.0  # Hz
t = np.arange(0, 40, 1 / fs)

# Ambient-like response on 3 channels (modes near 2.5 Hz and 6.0 Hz)
signal = np.column_stack(
    [
        np.sin(2 * np.pi * 2.5 * t) + 0.4 * np.sin(2 * np.pi * 6.0 * t),
        0.8 * np.sin(2 * np.pi * 2.5 * t + 0.2) + np.sin(2 * np.pi * 6.0 * t),
        0.5 * np.sin(2 * np.pi * 2.5 * t) + 0.7 * np.sin(2 * np.pi * 6.0 * t + 0.1),
    ]
)
signal += 0.05 * rng.standard_normal(signal.shape)

fdd = FDD(
    frequency_min=0.0,
    frequency_max=20.0,
    number_of_fft_points=512,
    num_svd_plots=3,
)

# Optional: inspect SVD lines before picking peaks
frequencies, eigenvalues, eigenvectors = fdd.compute_signal_svd(
    signal=signal,
    sampling_frequency=fs,
)

# Results are TypedDict mappings (use bracket access)
results = fdd.apply(
    signal=signal,
    sampling_frequency=fs,
    peaks_range=[[2.0, 3.0], [5.5, 6.5]],
)

print(results["frequencies"])       # ~2.54 Hz, ~6.05 Hz
print(results["complex_mode_shapes"].shape)  # (n_channels, n_modes)
print(results["real_mode_shapes"].shape)

Covariance SSI

from dynoma import CovSSI

cov_ssi = CovSSI(
    frequency_min=0.0,
    frequency_max=20.0,
    order_min=20,
    order_max=40,
    order_steps=2,
    time_lag=1.2,
    number_of_fft_points=256,
    num_svd_plots=3,
    continuous_mode=True,
)

results, dashboard = cov_ssi.apply(
    signal=signal,
    sampling_frequency=fs,
    optimized=True,
    shuffle=False,
)

print(results["frequencies"])
print(results["damping_ratios"])  # percent scale
# dashboard is None when continuous_mode=True

Public API

from dynoma import FDD, CovSSI, OmaAlgorithm, ModalIdentificationError, __version__

Signal layout is (n_samples, n_channels).

Requirements

• Python ≥ 3.10
• NumPy, SciPy, Pydantic, PyTorch

CPU-only PyTorch wheels are preferred for development (uv is configured for the official CPU index).

License

MIT — see LICENSE.