Assessment and Reduction of Uncertainties in Operational Modal Analysis (RUN-OMA) : Supplemental Data

The following list the folders and subfolders of the repository and a short summary of its contents. Also the naming conventions for the files in these folders will be stated.

estimations: contains subfolders with the data for the PolyUQ objects at the different stages of the uncertainty quantification (UQ) process.

• The subfolders follow the naming convention:
  {output}_{method}_{mode_no} where
  • output is one of :
    • d: damping ratio,
    • f: natural frequency,
    • scs: signal clarity score,
    • snr_db: signal-to-noise ratio,
    • sum_mc: modal reconstruction percentage
  • method is one of the OMA methods:
    • cf: poly-Reference Least Squares Complex Frequency (pLSCF)
    • sc: covariance-driven Stochastic Subspace Identification (SSI-cov)
    • sd: data-driven Stochastic Subspace Identification (SSI-data)
  • mode_no: is the number of the 13 vibration modes (0 … 12) of interest
• Each subfolder contains the following files (in the order of the UQ process:
  • polyuq_prop.npz: the propagated output samples
  • polyuq_sens.npz: the results of the sensitivity analyses
  • polyuq_imp.npz: the results and parameters of the Imprecision processing analyses
  • polyuq_{estimator}_inc.npz: the final results and parameters of the Incompleteness and Variability processing analyses, where estimator is the used statistical estimator and one of
    • avg: the average estimator
    • hist: the histogram estimator
    • cdf: the empirical cumulative probability distribution estimator

samples: contains the input and output databases and supplemental data of the uncertainty processing step, which was performed in parallel on a HPC cluster. The majority of intermediate result data (~20 TB) is not present in this repository but is stored on tape storage at the Bauhaus-Universität Weimar and can be supplied upon reasonable request.

• dm_oma_a.nc: The input database as a NetCDF file to be opened with the python module xarray
• dm_oma_a_out_pre_clust.nc: The raw output database as a NetCDF file to be opened with the python module xarray
• dm_oma_a_out.nc: The output database post vibration mode-clustering as a NetCDF file to be opened with the python module xarray
• mechanical_frf.dat: The frequency response function of the mechanical model that was used to generate output vibration data from wind field input forces
• mechanical.npz: the model data to be used for setting up the mechanical model

figures: contains all figures, both in PDF and PNG format, that have been derived from the result data. A selection of these will be published in the Dissertation of Simon Marwitz. The figures follow different naming conventions:

• imp_loo_accuracy{output}_{method}_{mode_no}: The histograms of the RBF interpolator leave-one-out validation errors (quality control of the Imprecision processing step).
• inc_{estimator}_{aggregate}_{output}_{method}_{mode_no}: Diagrams derived from the final Incompleteness processing results. For the definitions of estimator,output,method and mode_no see above. Additionally, method or mode_no may be all indicating a plot that contains multiple methods or vibration modes. Aggregate may be one of the Belief aggregation methods
  • foc: Plot of Focal intervals, no belief aggregation
  • pl: Plot of Plausibility functions derived from the Focal intervals.
• rbf_epsilon{output}_{method}_11: The interpolation error of the RBF interpolator with varying kernel shape parameter epsilon.
• sensi{output}_{method}_{mode_no}: Errorbar plots of the sensitivities of the respective output quantity with respect to all input variables
• sensi_stage{n}_{method}_: Combined plot of the sensitivities of all outputs of the respective stage of the OMA process model:
  • 23: stages 2 and 3 (Ambient Vibration Testing and Signal Processing)
  • 4: Final stage (Modal Analysis)

At the root level, the following files reside:

• example_reconstruction_validation_{method}: result data of the reconstruction error method for the three OMA methods
• polyuq_samp.npz: The input samples for the of the UQ process
• polyuq_samp_stage1.npz: The input samples for the of the UQ process with the first stage of the OMA process model