The Construction and Analysis of Prediction Model in Combining Taguchi Method and Neural Network to the Dying Process

Abstract

The purpose of this study is to find the optimum combination of processing parameters for achieving the target required. Taguchi method has been proposed in the research. The orthogonal array is used to determine the optimum conditions and significant factors together with the analysis of variance (ANOVA). In the experiment, the color strength (K/S) of fabrics was chosen to be the target characteristic, and the confirmation experiments were performed to verify the reproducibility of the experimentation. After three confirmation experiments, the signal-to-noise ratios (SN ratios) fall within 95% confidence interval. In addition, the K/S values of dyed fabrics in optimum condition are much closer to the target values. In conclusion, the significant factors influencing the dyeing results were used to construct the prediction system of back-propagation neural network combined with Taguchi method for achieving the accurate prediction ability. The results indicate that the percentage errors are all within 3.5.