Model prediction and numerical simulation on melt temperature distribution in process of polymer extrusion

Polymer dynamic extrusion is a complicated MIMO nonlinear system with time-varying, and the melt temperature is one of key parameters to measure and control. In order to improve the predicting accuracy of melt temperature distribution under the influence of multi-variable coupling, a ridge regression method based on Gaussian RBF (GRBF-RR) is presented. The model fulfill the nonlinear mapping and reconstruction of high-dimension feature space from multi-variable input samples, and is superior to least square method in eliminating potential multicollinearity of variables new-produced. The simulation results show that the GRBF-RR model makes a full characterization of internal laws of melt temperature in process of polymer dynamic extrusion. The numerical simulation based on the GRBF-RR model illuminates the coupling influence of input variables on the melt temperature distribution, and provides decision support for the optimization quality control of plastic product machining.