Computer modeling of the nonchromate treatment of aluminum alloys by neural networks

We analyze the thermodynamic characteristics and specific features of the course of reactions of oxometalate reduction on the aluminum surface and the properties of the final products (the specific electric resistance of oxides). The absence of a correlation between the ranged series of these parameters led to the necessity of using artificial intellect for modeling the processes of formation of conversion coatings. We obtained and visualized in 3D space the dependences of the corrosion resistance of D16 alloy on the oxometalate concentration, pH, the time of sealing, and test duration. Artificial neural networks based on multilayer perceptrons were used for predicting the corrosion resistance of conversion coatings. © Springer Science+Business Media, Inc., 2008.