WARPAGE PREDICTION IN PLASTIC INJECTION MOLDED PART USING ARTIFICIAL NEURAL NETWORK

The main objective of this paper is to predict the warpage of a circular injection molded
part based on different processing parameters. The selected part is used as spacers in automotive,
transmission, and industrial power generation industries. The second goal is facilitating the setup
of injection molding machine without (any) need for trial and error and reducing the setup time. To
meet these objectives, an artificial neural network (ANN) model was presented. This model is
capable of warpage prediction of injection molded plastic parts based on variable process
parameters. Under different settings, the process was simulated by Moldflow and the warpage of
the part was obtained. Initially, the effects of the melt temperature, holding pressure and the mold
temperature on warpage were numerically analyzed. In the second step, a group of data that had
been obtained from analysis results was used for training the ANN model. Also, another group of
data was applied for testing the amount of ANN model prediction error. Finally, maximum error of
ANN prediction was determined. The results show that the R-Squared value for data used for
training of ANN is 0.997 and for the test data, is 0.995.