Constitutive modeling of Ta-rich particle reinforced Zr-based bulk metallic composites in the supercooled liquid region by using evolutionary artificial neural network

Abstract

In this study, a series of in-situ Ta-rich particle reinforced Zr-based bulk metallic glass composites were successfully fabricated by arc-melting copper-mold spray casting. The effects of Ta content on the room temperature plasticity, compressive strength and thermoplastic formability were studied. (Zr55Cu30Al10Ni5)94Ta6 showed good comprehensive performance, and it was selected to systematically study the deformation behavior in the supercooled liquid region. Different from the strain softening after stress overshoot in bulk metallic glass, the composites showed work hardening in the late stage. Some classical constitutive models cannot accurately describe these phenomena. The back-propagation artificial neural network optimized by particle swarm optimization and genetic algorithm was used to establish the constitutive model. The particle-swarm-optimization back-propagation network with the optimal topology showed high accuracy and good generalization ability. The results predicted with this model were consistent with the experimental data, providing a powerful approach for describing the hot-deformation behavior of these Zr-based bulk metallic glass composites in the supercooled liquid region.