Study on design optimization of  cream box   injection mold using hybrid orthogonal table and artificial neural network

Abstract:

In injection molding, if the die design is unsuitable, defects such as welding line, surface burning, shrinkage, residual stress and warpage on the surface of the plastic product are present.
When the process is determined by the test injection and the design of the die in an empirical way, the quantitative parameters are not known, so this method cannot improve the quality of the injection product and ensure high production efficiency. However, computer and software are now progressing rapidly, with the development of specialized CAD/CAM/CAE applications for machine design, analysis and simulation.
In this paper, based on the analysis of the cream-box injection product by the injection molding analysis program (Moldflow), an injection-mold design optimization was carried out by combining the method by BP neural network with the modern mathematical method. 
The design optimization of the product thickness, the size and number of guide grooves, the flow path and the cooling system of the die with minimum bending strain were carried out, and the bending strain of the injection molded product was estimated and predicted under the optimum conditions.