Reversing design on combined parameters of liquid-solid extrusion process based on the predictive model using hybrid GA-BP algorithm

In order to control deformation uniformity of composite in the forming process of liquid-solid extrusion and reduce inner damaging defects of products. Based on artificial neural network (ANN) technique and genetic algorithm (GA), the nonlinear mapping relation between design variables and objective function was proposed and established by the modified GA-BP algorithm. The simulation results of FEM called virtual samples were selected as the network's training samples. By training the sample, the knowledge base of the multi-parameters for the liquid-solid extrusion was set up. Comparing with the experimental results, the largest relative error between the actual output value of the network and the experimental data is 0.79 percent. It proves that the forecast model established using GA-BP hybrid algorithm has a higher accuracy. The influences of main process parameters and structures parameters had been studied on the deformation uniformity using the predictive function of the model. They are good instructions for the design and optimization of the liquid-solid extruding composites process.