DYNA JOURNAL ENGINEERING

Key words:

Moldeo por inyección de plásticos, Esfuerzo de tensión, Poliamida-6, Superficie de Respuesta, Red Neuronal de Retro-propagación, Red Neuronal de Regresión Generalizada, Programación no lineal, Plastic Injection Molding, Tensile stress, Polyamid-6, Response Surface, Backpropagation Neural Network, Generalized Regression Neural Network, Nonlinear programming

Article type:

ARTICULO DE INVESTIGACION / RESEARCH ARTICLE

Section:

RESEARCH ARTICLES

The objective of this research is optimizate tensile stress of injection molded parts of polyamid-6 to establish process conditions that maximize tensile strength of parts in a real industrial process. The methodology consisted in development of essays based on I-optimal experimental design in order to get a data base. During experimentation four parameters were considered as inputs: injection holding pressure, injection time holding, % wt virgin material and % wt recycled material. Measurement of máximum tensile stress in parts was made according to ISO 527-1 standard.
Three models were developed by the techniques Response Surface Metodology, Back Propagation Neural Network and Generalized Regression Neural Network (with Radial Basis Function) to predict parts máximum tensile stress. Finally, the best model (lowest forecasting error) was optimized by Trust Region Method Based on Interior Point Techniques for Nonlinear Programming to maximize tensile strength.
Was concluded that this proposed methodology is capable for modeling the injection molding process with low error and for stablish process conditions to obtain the máximum tensile stress on molded parts.
Keywords: Plastic Injection Molding, Tensile stress, Polyamid-6, Response Surface, Backpropagation Neural Network, Generalized Regression Neural Network, Nonlinear programming.