Warpage Analysis And Process Optimization Of Thin-walled Plastic Parts Injection Molding

Thin-walled molding injecting is a new technology for thin-walled products that based on the traditional injection molding. The thin-wall design can reduce the weight and external dimension of the products, benefit integrated design and assembly, shorten production molding cycles, economize material and lower costs, etc. But as the thickness of plastic parts become thinner, It is much more difficult to fill the cavity because the melt is frozen very quickly. Warpage is the main defect and can lead many final assembly failures in manufacturing process. Plastic injection is non-linear and flexible multi-factors manufacturing process. Due to many factors involved, their impacts on the final quality are different. Therefore, analyzing the warpage of the plastic parts and optimizing their process parameters, is a very effective measure to improve the production quality.In this paper, firstly provides a brief overview of the thin-walled plastic injection molding technology, describes the generation mechanism of warpage. And then analyze the factors that affect the warpage, and that for reducing warpage of thin pieces of multi-kinds of measures, and then research methods of warp are discussed and determined.On this foundation, taking a thin-walled plastic part for example, when the warpage is treated as optimization objective, the Taguchi design and CAE technology were applied to research the effects of mold temperature, melt temperature, injection rate, hold pressure and packing time on part's quality. Two optimum parameter combinations for shrinkage and warpage were obtained by ANOM. And then, the degree of each process parameter's effect on quality was obtained by ANOVA. The process parameter combination giving attention to both shrinkage and warpage was selected by the synthetically weighted method of point.Finally, taking data from CAE as samples; the BP neural network of warping-shrinkage prediction model is established by designing the network structure and selection of learning algorithm. Comparing with the results of the network and the simulation and error analysis, showed that trained neural network model can correctly predict the quality of the products, and Significantly reduce the number of numerical simulation. Achieve the optimization of the injection process conditions, shorten production time and improve part quality.