A study on volumetric shrinkage of injection molded parts based on neural networks

Shrinkage of injection molded parts is caused by characteristics such as differential cooling or differential orientation affected by variables such as materials, geometry of a part or mold, and processing conditions. The effect of processing conditions on volumetric shrinkage is emphasized in this paper, and the relationship between processing conditions and volumetric shrinkage of an injection molded part is analyzed by a neural network constructed using the learning data extracted from simulation software. A plastic gear made from an engineering plastic is used for the study, and the range of experimental conditions is decided by the orthogonal table based on the design of experiments within the processing windows. A nonlinear regression model is formulated using the 3125 data points obtained from a neural network, and the optimal processing conditions to minimize the volumetric shrinkage of an injection molded part are calculated by the application of a recursive quadratic algorithm, and it is compared with those of the neural network and simulation.