An inverse model for injection molding of optical lens using artificial neural network coupled with genetic algorithm

详细信息    查看全文

• 作者：Kuo-Ming Tsai ; Hao-Jhih Luo
• 关键词：Inverse model ; Artificial neural network (ANN) ; Genetic algorithm (GA) ; Injection molding ; Optical lens ; Taguchi method
• 刊名：Journal of Intelligent Manufacturing
• 出版年：2017
• 出版时间：February 2017
• 年：2017
• 卷：28
• 期：2
• 页码：473-487
• 全文大小：
• 刊物类别：Business and Economics
• 刊物主题：Production; Manufacturing, Machines, Tools; Control, Robotics, Mechatronics;
• 出版者：Springer US
• ISSN：1572-8145
• 卷排序：28

文摘

This study combined the artificial neural network (ANN) with a genetic algorithm (GA) to establish an inverse model of injection molding for optical lens form accuracy. The Taguchi parameter design was used for screening experiments of the injection molding parameters, and the significant factors influencing lens form accuracy were found to be mold temperature, cooling time, packing pressure, and packing time. These significant factors were used for full factorial experiments, and the experimental data then were used as training and checking data sets for the ANN prediction model. Finally, the ANN prediction model was combined with the GA to construct an inverse model of injection molding. Lens form accuracies of 0.5, 0.7, and \(1\,\upmu \hbox {m}\) were taken as examples for validation, and when the error of the set lens form accuracy target value was within 2 % there were 26, 17, and six sets of the injection molding parameters, respectively, that met the desired form accuracy obtained by using the inverse model. The result indicated that the proposed strategy was successful in identifying process parameters for products with reliable accuracy. In addition, using the GA as a global search algorithm for the optimal solution could further optimize the Taguchi optimal process parameters. The validation experiments revealed that the form accuracy of the lens was improved.