Sequential design of an injection molding process using a calibrated predictor

• Autores: Po Hsu Allen Chen, M. Guadalupe Villarreal Marroquín, Angela M. Dean, Thomas J. Santner, Rachmat Mulyana, José M. Castro
• Localización: Journal of quality technology: A quarterly journal of methods applications and related topics, ISSN 0022-4065, Nº. 50, 3, 2018, págs. 309-326
• Idioma: inglés
• Enlaces
  • Texto completo
• Resumen

  • This article optimizes an injection molding process using an efficient sequential design methodology. The goal is to set the process control variables to minimize the shrinkages of a selected collection of injection molded parts. This multiobjective optimization problem is solved by finding those process control variable settings that are Pareto minimizing values (i.e., process settings for which none of the shrinkages of the parts can be decreased by an alternative process setting without increasing the shrinkages of other parts). The sequential design uses an expected improvement criterion to guide updates. The shrinkages are estimated by a calibrated predictor of the process mean shrinkage. The calibration is based on observations of the manufacturing process supplemented by computer runs of a commercial simulator code that mimics the manufacturing process.