Data Processing (VRM)¶
The data obtained from the multi-fidelity simulation software (VRM) consists of node deviations, these need to be processed and mapped to the corresponding voxel location as voxel features using the mapping index for the nominal cloud of point (comes with the downloaded files for the case study with 64*64*64 resolution, different resolution mapping index can be generated using the voxelization utility).
Contains classes and methods to process the VRM data and convert it to the format as required by the 3D CNN model
-
class
dlmfg.core.data_import.GetTrainData[source]¶ GetTrainData Class (No initialization parameter)
-
data_convert_voxel_mc(vrm_system, dataset, point_index, kcc_data=Empty DataFrame Columns: [A] Index: [])[source]¶ data converts the node deviations to voxelized output
- Parameters
vrm_system – Object of the VRM System class
dataset (list (required)) – list of concatenated dataset consisting of x,y,z deviations for each node
point_index (numpy.array [nodes*3] (required)) – mapping index
kcc_data (numpy.array [samples*kcc_dim] (required)) – Process parameter data
- Returns
input_conv_data, voxelized data for model input
- Return type
numpy.array [samples*voxel_dim*voxel_dim*voxel_dim*3]
- Returns
kcc_data_dump, process/parameter data for model output
- Return type
numpy.array [samples*kcc_dim]
- Returns
kpi_data_dump, KPI data (if any) for each sample, convergence flag (convergence of simulation model) is always the first KPI
- Return type
numpy.array [samples*kpi_dim]
-
data_import(file_names, data_folder)[source]¶ data import used to import all files within the given folder and concatenate them into one dataframe
- Parameters
file_names – List of the input files
data_folder (str (required)) – data folder name
- Returns
dataframe of concatenated data from each file within the list
- Return type
pandas.dataframe [samples,point_dim]
-