| Injection Molding is classified as one of the most flexible and economical manufacturing processes with high volume of plastic molded parts. Causes of variations in injection molding are related to the vast number of factors acting during a regular production run, which directly impacts the quality of final products. Injection molding companies are interested on determine when a common defect such as warpage might be present on their finished products. Five critical process parameters are employed in this study for being considered to be critical process parameters that have a great impact on the warpage of molded parts. Therefore, melt temperature, mold temperature, packing pressure, packing time, and cooling time are used as key variables of the injection molding process. Injection molding process simulations are carried out by using Autodesk Mold Flow Insight software. This study aims to design an intelligent system based on artificial neural networks to predict quality defects in injection molding. Artificial neural networks are mathematical models that can accept a large set of inputs and learn from training samples. Artificial neural networks are categorized into two groups: feedforward neural networks and recurrent neural networks. Recurrent neural networks are able to deal and understand dynamic systems; it is assumed that recurrent neural networks may work to predict when a defect will be present in the injection molding process. In this research, a combination of two recurrent architectures are evaluated namely Elman network and Jordan network. Using this type of architecture, it is assumed that the network adjusts in all its layers. Results show the proposed methodology works well in prediction tasks, overcoming those results generated by common neural networks. |
