Machine learning-enhanced constitutive modeling of hot deformation in FeCr1.2Ni1.5Al multi-principle element alloys: A denoising neural network approach

This work integrated artificial intelligence (AI) and hot processing maps to offer a powerful framework for optimizing hot deformation processes of FeCr1.2Ni1.5Al MPEAs. Benefiting from superior noise handling, the Wavelet Neural Network (WNN) model was innovatively proposed to construct constitutive model, which had been demonstrated high accuracy and reliability in predicting flow stress as a function of temperature, strain rate, and strain. The wavelet decomposition allowed WNN to analyze stress-strain behavior at different frequency levels, effectively identifying localized and global trends in mechanical response. Using the constitutive models, we constructed hot processing maps to identify optimal hot working conditions at temperatures of ( > 1050 ∘ C) with strain rates of ( < 0.01 s − 1 ), revealing safe processing windows. The deformed microstructural results provided detailed insights into grain orientation, grain boundary characteristics, local misorientations, and texture development, further elucidating the underlying deformation mechanisms. The higher temperatures enhanced dynamic recrystallization (DRX) and dynamic recovery (DRV), producing fine equiaxed grains, while lower temperatures led to deformation bands and high dislocation densities. The higher strain rates promoted deformation-dominated microstructures, while lower strain rates favored DRX and grain refinement. The findings provided a comprehensive understanding of the microstructure-property relationships under different deformation conditions, paving the way for the development of advanced materials with tailored properties for high-performance applications. • Hot deformation behavior of FeCr1.2Ni1.5Al MPEA exhibited high flow stress at elevated temperatures. • Wavelet Neural Network with denoising technique was innovatively proposed to establish high-precision constitutive model. • The optimal processing parameters were derived from hot processing maps. • Hot deformation mechanisms were discussed with the effects of temperature and strain rate.