Abstract: In order to produce insert-gear plastic toys more effectively, an injection mold with 144 cavity was designed using natural balance method. The finite element model was reasonably simplified and the flow and warpage in the molding was analyzed by Moldflow software. Focused on the welding marks and warpage and other defects, a BP artificial neural network model was established, the process parameters were trained, the volume shrinkage and the total amount of warpage were predictively analyzed by the BP neural network algorithm. The optimal combination of process parameters after training was applied to the injection molding, the weld mark and warpage were reduced significantly.

Key words: injection molding, Moldflow, gear toy, BP neural network, optimization, analysis