New Approach in Predicting Aluminum-Alloy Properties Using Neural Networks
In this study the impact of alloying elements on hardness, for different series of Aluminum alloys, was analyzed
and studied using a mathematical modeling tool based on Artificial Neural Networks (ANN). This new model was developed
as multi-layer feed forward network with a hidden layer that can approximate any complex function and back-propagation
learning algorithm. Thus, it is possible by using this model to predict properties of Aluminum alloys, in particular hardness,
based on certain punch of inputs. The latter includes the weight percentage of the composition contents of various metals
such as Cr, Cu, Fe, Mg, Mn, Si, Ti, Zn, and the Aluminum. In the current research work, several runs were made using this
newly developed model, and published results of previous experimental research, available in open literature. The obtained
results here were found in full agreement with those reported by other researchers. Thus, it is deemed that such model could
help researchers and industries in designing the desired Aluminum alloy and predicting its hardness at no cost. Equally
important is relatively high accuracy of obtained results.
Keywords - Aluminum alloy, Alloying elements, hardness, Neural Networks.