ANN and GA-based process parameters optimization for plastic injection molding

• Main Authors: Yang-Chih Fan, 范揚志
• Other Authors: Wen-Chin Chen
• Format: Others
• Language: zh-TW
• Published: 2006
• Online Access: http://ndltd.ncl.edu.tw/handle/77189609008655504101

碩士 === 中華大學 === 科技管理研究所 === 94 === Abstract In plastic injection modeling industry, every product should be set up by its own process parameter due to its versatility. In the past, engineers relied on production experience and intuition to control process parameters and product quality for injection molding. However, the high mobility of personnel and the large variation in manufacturing make the traditional trial-and-error technique ineffective. This study proposes artificial neural networks (ANN)-based system, which employs Taguchi Orthogonal Arrays to implement the experiment on process parameters of injection molding system (IMS) and identifies a better set of initial process parameters, to construct a quality predictor through the training and testing of back-propagation neural network (BPNN). In addition, the proposed genetic algorithm (GA) combines the quality predictor to explore the optimal process parameters. The experiment results show that the proposed model facilitate achieving the requirement of quality characteristics (i.e. length and weight), reducing the number of die testing and making the production more efficient, economical and convenient in injection process.