A novel optimization system applied in injection molding of a LED lighting module

• Main Authors: Tai, Yi-Chia, 戴鎰家
• Other Authors: Chen, Wen-Chin
• Format: Others
• Language: zh-TW
• Published: 2013
• Online Access: http://ndltd.ncl.edu.tw/handle/wy5uj5

博士 === 中華大學 === 科技管理博士學位學程 === 101 === In the injection molding process of a LED lighting modulus, trial-and-error processes and the design of experiments are frequently employed to determine initial process parameter settings, which depend on the engineers’ experience and intuition. However, since many different parameters could influence a finished product and a high level nonlinear relationship stands between each parameter, it takes a large number of manpower, devices, and other expenses to figure out a better combination of process parameters. Thus, this study presents a novel optimization system for injection molding with multiple performance characteristics through data mining and analysis to effectively determine the optimal process parameter settings. The quality characteristics of the LED lighting modulus can be categorized into the beam angle and the luminous intensity. The control factors for the process are mold temperature, melt temperature, injection velocity, packing pressure and VP switch. The proposed parameter optimization system is divided into two stages. In the first stage, the Taguchi method is employed to conduct signal-to-noise (S/N) ratio optimization. Taguchi orthogonal array experiments are performed, and then the experimental data are trained and tested by back-propagation neural networks to create a S/N ratio predictor and a quality predictor. In addition, the S/N ratio predictor is combined with genetic algorithms (GA) to obtain the process parameter combination on maximum S/N ratio for both beam angle and luminous intensity. As a result, the quality variance could be reduced to minimum. In the second stage, optimization of quality characteristics is carried out. Analysis of variance (ANOVA) is employed to determine the control factors of numerical analysis; the afore-mentioned quality predictor and S/N ratio predictor along with hybrid GA-PSO is to implement the global search and to draw close to the target of specification, and available to generate the most stable and low-defective ratio product. The proposed novel optimization system can create the best process parameter settings which can not only be more robust and meet the dimension specification of a LED lighting modulus, but also enhance the stability of injection process and its product quality characteristics.