Application of artificial neural networks for modelling correlations in titanium alloys

Abstract

This paper is dedicated to the application of artificial neural networks (ANN) in titanium alloys research, including: (i) time–temperature transformation (TTT) diagrams for titanium alloys; (ii) correlation between processing parameters and properties in titanium alloys and γ-TiAl-based alloys; (iii) fatigue stress life diagrams for Ti–6Al–4V alloy; (iv) corrosion resistance of titanium alloys. For each particular case, appropriate combination of inputs and outputs is chosen. Standard multilayer feedforward networks are created and trained using comprehensive datasets from published literature. Very good performances of the neural networks are achieved. Different effects are modelled, among which are: (i) influence of the alloying elements on the transformation kinetics in titanium alloys; (ii) influence of the processing parameters, alloy composition and the work temperature on the mechanical properties for titanium alloys and titanium aluminides; (iii) influence of the microstructure, temperature, environment, surface treatment and the stress ratio on the fatigue life. The artificial neural networks models are combined with computer programmes for optimisation of the inputs in order to achieve desirable combination of outputs. Graphical user interfaces are developed for use of the models. These models are convenient and powerful tools for practical applications in solving various problems in titanium alloys.