GPC CURVES (*)

TREATMENT OF COMPLEX DATA

Observation of time curves, spectrums. 

To build relevant indicators for the GPC analysis, from curves collected on a product or a production cycle.

The principle of this module is to define indicators best reflecting:

  • • Variability between curves.
  • • Different disruptions.

GPC Curves module comes into the GPC approach of complex situations.

In fact, in some industries, it is not enough to collect individual scalar measures: each product or elementary operation process is associated to one or several curves each composed of hundreds, even thousands of points:

  • Time curves (plastic processing, semiconductor, combustion engine…)
  • Spectral curves (sound or vibrational spectrums for rotating system, infrared spectrums in chimical or biological industries…).

The classical approach is based on a process expertise feedback. It is about choosing some curves characteristics in order to allow a differenciation between normal operation of the system and disturbed operation (value on a point, peak position, average value on a period…). These characteristics are monitored by means of SPC control cards.  

The GPC approach of this situation is broken down in two phases:

  • Construction of “optimal” from statistical analyses of a learning play of curves. It complete the more relevant “expert” indicators.
  • Multivariate GPC monitoring of the defined indicators.

In several industrial situations, this method allows automatic and economic 100% control in replacement control by sampling eventually, descructive.

(*) The module is not currently accessible on the GPC Box service in the Cloud. It can be used as part of an accompaniment study by GPC System.

Examples of application:

ANALYSIS OF TIME CURVES:

  • •Plastic injection.
  • Semiconductor.
  • Combustion engines…

ANALYSIS OF SPECTRUMS:

  • •Chemical industry.
  • •Engines, generators.
  • •Biological systems…

This example relates to plastic processing.

For each product made, several temperature or pressure curves are collected in the molding or in the mold.

The GPC analysis applied to only two indicators built by GPC Curves on a unique type of curve, allows to perfectly distinguish the two types of wrong products.