A data-driven material constitutive equation reverse mining method

Abstract

This invention belongs to the interdisciplinary field of plasticity mechanics and intelligent algorithms, specifically a data-driven method for reverse mining of material constitutive equations. It involves obtaining mechanical response data of plastic materials under different working conditions through experimental testing, supplementing extreme working condition data and control parameters corresponding to the initial material constitutive equation through numerical simulation, forming a multi-source mapping dataset of stress state-fracture response; calibrating the parameters of the initial material constitutive equation using the multi-source mapping dataset; generating plasticity response datasets in batches based on the calibrated initial material constitutive equation through parameter traversal and working condition expansion; configuring the core parameters of a Transformer-based symbolic regression model, then training the model using a training set to output several candidate equations; and evaluating the accuracy, complexity, and mechanical rationality of the candidate equations using a validation set to select the optimal material constitutive equation.

Images