Multiple Quality Characteristics Optimization of Precision Injection Molding for LCD Light Guide Plates

Abstract

A light guide plate is a key component that enables backlight modules to provide sufficient light source for liquid crystal displays (LCDs). Hence, this study aims to investigate the relationship between precision injection molding process parameters and light guide plate's quality characteristics. Taguchi method and grey relational analysis are applied to determine the optimal processing conditions for light guide plates and back-propagation neural network (BPNN) is used to establish a quality prediction system for molded light guide plates. First, the study determines the quality characteristics of light guide plates and uses Taguchi method to plan experiment. Nonetheless, since the Taguchi method is mostly used to optimize single quality characteristic, which often fail to represent process parameter optimization of the overall quality, this study takes advantage of grey relational analysis and integrates multiple qualities to determine the optimal process conditions for light guide plates. The Taguchi method is chosen to design the learning parameters for the BPNN, which is free from the drawbacks of the traditional trial-and-error method and speeds up network convergence to determine preferable combinations of learning parameters. As the experiments reveal, the prediction system established in this study is effective in accurate prediction of the qualities of light guide plates.