Applications of artificial intelligence methods for modelling of solidus temperature for hypoeutectic Al-Si-Cu alloys

This paper presents the application of neural networks for prediction of the solidus temperature of various hypoeutectic Al-Si-Cu casting alloys cooled with different Cooling Rates (CR). Knowledge of solidus temperature allows the prediction of a variety of metallurgical characteristics, that is, melt treatment, casting temperature and solidification range. Currently, the literature reports only one equation for determination of solidus temperature for hypoeutectic aluminium alloys. This paper presents computational algorithm, comparison with different models' predicted solidus temperature and influence of alloying elements on solidus temperature. The results of this investigation show that there is a good correlation between experimental and calculated dates and the neural network has great potential in modelling of solidus temperature of Al-Si-Cu alloys. The worked out model can be applied in the computer system for calculating of chemical composition and CR influence on the solidus temperature of Al-Si-Cu alloys. [Received 22 November 2006, Accepted 15 January 2007]