A study on blending polyethylene terephthalate with titanium dioxide particles in melt spinning process parameter optimization

This study blended polyethylene terephthalate and titanium dioxide nano-particles to produce titanium dioxide-containing (TiO2) functional fiber, and discussed its ultimate tensile strength, elongation and modulus of resilience using the Taguchi orthogonal arrays for experiment planning. The main effect analysis and analysis of variance theory were applied to obtain the single attribute optimum parameter, and Data Envelope Analysis and Principle Component Analysis were conducted to determine the combination of the optimum process parameter levels. Finally, the confidence interval was calculated by confirmation test, and a Back-propagation Neural Network was applied to construct the prediction system for melting spinning process parameters. After the learning and training of the network, the prediction error rate of the prediction system was below 5%.