The Predicting Model of Superplastic Property of Aluminum Bronze and the Superplastic Extrusion Test

摘要

The superplastic properties of aluminum bronze were studied by way of artificial neural network (ANN). The predicting model was established using Levenberg-Marquardt algorithm. And it was perfected by studying the stylebook of superplastic tension such that the superplastic forming parameters were optimized. According to the parameters, the experiment of superplastic deformation of the bearing cage was performed. It is shown that the model reflected well the relationship between superplastic properties of aluminum bronze and superplastic tension conditions. The relative error between the expected values and the predicted outputs of the network is less than 8.5 %, which meets perfectly the demands of superplastic deformation of aluminum bronze. Moreover, the superplastic experiments of the bearing cage of aluminum bronze show that it is feasible to produce the bearing cage using superplastic extrusion process. This extrusion process has remarkable economic benefits as well.