Comparison of injection molding process windows for plastic lens established by artificial neural network and response surface methodology

Abstract

This study used artificial neural network (ANN) and response surface methodology (RSM) to obtain the lens form accuracy prediction model. Moreover, it established the operating parameter region that satisfies the designated lens form accuracy at a minimum cost. Finally, experimental verification and accuracy comparison were conducted. This study used the Taguchi method for parameter screening experiment in the injection molding process. The obtained significant factors that influence form accuracy of the lens were mold temperature, cooling time, and packing time. Those significant factors were used for full factorial experiment, for adjusted experiment using composite central design method, and to establish the process window. According to the results, the process window for injection molding using an ANN to establish cooling time and packing time was a high-order, irregular shape, whereas the one established by a RSM was oblique oval. The optimal form accuracy of the lens obtained from ANN model was better than that from RSM model. The result of the experiment indicated that the process window of the injection molding process for optimal form accuracy obtained from both ANN and RSM models is quite consistent. In addition, a case study for a form accuracy of 0.5 μm was discussed. The process window established by ANN had a better accuracy and wider range than that by RSM.