Prediction of core deflection in wax injection for investment casting by using SVM and BPNN

Abstract

Core shift is a common basic problem during the manufacture of those thin-walled hollow casting parts in investment casting. The core deflection during wax injection is the main reason for exceeding tolerance in final cast. The wax melt may advance asymmetrically along both sides of the core in wax injection process for a thin-walled hollow complex part, which causes the core to shift, resulting in non-uniform wall thickness of the wax pattern. The simulation results and experimental measurements of core deflection on the wax pattern for a hollow pump body were studied. The SVR (support vector regression) has been used to predict the core deflection for the wax pattern. Packing pressure, injection temperature and injection velocity were chosen as main process parameters. The core deflection of the wax pattern is most influenced by the melt temperature because of larger stress distribution acting on the cores in the viscous state at low injection temperature. SVR and BPNN (back-propagation neural network) were used to establish the prediction model; SVR has a better predictive ability compare with BPNN.