# THE CONSTRUCTION AND ANALYSIS OF A PREDICTION MODEL FOR COMBINING THE TAGUCHI METHOD AND GENERAL REGRESSION NEURAL NETWORK FOR INJECTION MOULDING

> Title: THE CONSTRUCTION AND ANALYSIS OF A PREDICTION MODEL FOR COMBINING THE TAGUCHI METHOD AND GENERAL REGRESSION NEURAL NETWORK FOR INJECTION MOULDING Page Range: p.823-829 Author(s): Chung-Feng Jeffrey Kuo; Yung-Chang Li; Yi-Shiuan Wu File size: 154K Download the pdf (subscribers only) Buy...

Canonical URL: https://polymerjournals.com/the-construction-and-analysis-of-a-prediction-model-for-combining-the-taguchi-method-and-general-regression-neural-network-for-injection-moulding/
Published: 2005-08-01T00:00:00+00:00
Categories: Polymers and Polymer Composites

**Title: THE CONSTRUCTION AND ANALYSIS OF A PREDICTION MODEL FOR COMBINING THE TAGUCHI METHOD AND GENERAL REGRESSION NEURAL NETWORK FOR INJECTION MOULDING**
 Page Range: p.823-829
 Author(s): Chung-Feng Jeffrey Kuo; Yung-Chang Li; Yi-Shiuan Wu
 File size: 154K
 ![](http://dload.gif) [Download the pdf](http://journals.asp?Page=111&JournalDownload=ON&JournalType=ppc&JournalIssue=ppc13-8&JournalID=100256&JIP=) (subscribers only)
 ![](http://handbuy.gif) Buy the pdf (non-subscribers)
 Journal: Polymers and Polymer Composites
 Issue Year: ppc
 Volume: 13
 Issue No: No. 8

**Abstract**
 In order to avoid dimensional errors after the injection moulding process, processing parameters are established to effectively provide dimensional control. In this work, particular attention is given to reducing the dimensional error in the hexagonal screw diameter when processing engineering plastics such as PEEK, and to also improve the quality of the polymer. Factor levels were chosen according to the chosen quality characteristics, and a prediction model for the injection process was constructed using the Taguchi quality method and a general regression neural network. The optimum conditions determined by the Taguchi method could be modified by the neural network. It was thus demonstrated that the injection moulded product could achieve reduced dimensional errors by adjusting the factor levels using this combined approach. 11 refs.

---