Test parameters optimization for constrained spray forming of aluminum alloy based on artificial neural network

Abstract

Spray deposition with following continuous extrusion (SD-CE) forming technique is a novel technology that combines spray forming and continuous extrusion. Optimization of test parameters for spray deposition is an important part of SD-CE. In this study, Al-20Si alloy was produced by spray forming at different melt temperature and gas pressure, and obtained grain diameter of 8 group primary silicon phase. Based on the experimental results, an Artificial Neural Network (ANN) with single hidden layers composing of 10 neurons was employed to simulate optimizing of test parameters for spray deposition. The inputs of the model are melt temperature and gas pressure. The output of the model is grain diameter. Finally, the minimum relative error of grain diameter is 0.09%, the maximum relative error is 8.38%, and error majority concentrate within 3.80%, the average absolute relative error(AARE) is 1.04%, R is 0.097, the error is small. The optimal test parameters for spray deposition are melt temperature(829 °C) and gas pressure(0.2 MPa). The results indicate that the ANN model is an easy and practical method to optimize the test parameters for spray deposition of Al-20Si alloy. Thereby this model is a useful reference for optimizing the test parameters of SD-CE