Key Measurement Characteristics (KMCs)

  • KMCs are useful is visualizing which areas need to be measured to determine the value of the process parameter. This is useful in terms of determining the optimal sensor layout and in assigning areas a priority when partial measurements needs to be conducted due to system constraints such as cycle time.

_images/kmc1.png

Fig 5: KMC for pin hole translation in x-direction

_images/kmc2.png

Fig 6: KMC for pin hole translation in y-direction

_images/kmc3.png

Fig 7: KMC for rotation around pin hole

Feature importance of ensemble models such as Gradient Boosted Trees or Random Forests is used in determining KMCs for each KCC

dlmfg.kmc_gen.kmc_model.kmc_model_build(tree_based_model, point_data, selected_kcc, kcc_name, split_ratio=0.2, save_model=0)[source]

kmc_model_build function inputs model_type and data to generate KMC for each KCC

Parameters
  • tree_based_model (str (required)) – Type of model to be used for feature importance

  • point_data (numpy.array (samples*nodes) (required)) – input data consisting of node deviations

  • selected_kcc (numpy.array (samples*1) (required)) – output data consisting of selected process parameter/KCC

  • kcc_name (str (required)) – unique identifier for the KCC

  • split_ratio (float) – test data split

  • save_model (int) – Save model flag, set to 1 to save model

Returns

filtered_nodeIDs_x, node ids for which x-deviation is significant given the kcc

Return type

numpy.array [kmcs*1]

Returns

filtered_nodeIDs_y, node ids for which y-deviation is significant given the kcc

Return type

numpy.array [kmcs*1]

Returns

filtered_nodeIDs_z, node ids for which z-deviation is significant given the kcc

Return type

numpy.array [kmcs*1]