Mathematical modelling of the physical and mechanical properties of nano-Y2O3 dispersed ferritic alloys using evolutionary algorithm-based neural network

• We develop artificial neural network models for nano-Y2O3 dispersed ferritic alloys
• Model predicts physico-mechanical properties of alloys for unknown Cr % and temp.
• Artificial neural network model was trained using genetic algorithm (GA).
• Error statistics demonstrated that the models are generalized and satisfactory.
• Results show that GA-based neural network performed better than other models.

This paper demonstrates the effects of chromium content and consolidation temperature on nano-Y2O3 dispersed ferritic alloys 83.0Fe–13.5Cr–2.0Al–0.5Ti (alloy A), 79.0Fe–17.5Cr–2.0Al–0.5Ti (alloy B), 75.0Fe–21.5Cr–2.0Al–0.5Ti (alloy C) and 71.0Fe–25.5Cr–2.0Al–0.5Ti (alloy D) each containing 1% Y2O3 (all in wt%) using mathematical modelling. An artificial neural network model is developed for the (a) physical properties (density and porosity) and (b) mechanical properties (hardness, Young's modulus, and compressive strength) of alloy. A three-layer neural network model was developed, and the genetic algorithm (an evolutionary algorithm) is applied for training the neural network. The results reveal that the proposed genetic-algorithm-based model performed better than the traditional neural network model and the linear regression model for modelling of the physical and mechanical parameters of chromium alloys. The simulated maps of the physical and mechanical properties of alloys were generated against the processing parameters, i.e., consolidation temperature and Cr (wt.%). The developed model can be used for optimization of the mechanical alloying process for synthesis of nano-Y2O3 dispersed ferritic alloys.

Figure optionsDownload as PowerPoint slide