基于纳米压痕试验的塑性本构参数反演方法研究

Nanoindentation technology is an essential experimental technique for examining the charac-teristics of biomaterials, thin films, and coatings. The method for determining the hardness and Young's modulus of materials using the load-displacement curve of an indentation test is rather mature at the mo-ment, but obtaining the plastic constitutive characteristics of materials remains a challenge. In this paper, a reverse analysis method for the plastic constitutive Parameters of M50NiL bcaring steel with the surface carburized by vacuum chemical heat treatment is establishcd. As the depth changes along the surface car-burization direction, these Parameters based on the Johnson-Cook plastic constitutive model are obtained. Finite element calculations are first performed with various plastic constitutive Parameters, which provide datasets for feeding the neural network. The trained neural network with acceptable precision can then be used to replace the complicated and time-consuming finite element computations to connect constitutive Parameters with indentation load-displacement curves. Knowing the constitutive parameters, the neural network can successfully Output the characteristics of matching indentation load-displacement curves, allowing the genetic algorithm to be used in reverse analysis. The genetic algorithm generates and evolves a popula-tion of candidate constitutive parameters over generations for a specific nanoindentation load-displacement curve from an experiment. These parameters are assessed using a fitness function, which determines how well they can contribute to a curve that is in good agreement with the target curve. The fitness function in the algorithm is based on the previously trained neural network to rapidly transfer the candidate constitutive parameters into load-displacement curve features, which should be compared with the target curve fea-tures to assess the candidate Solution. After evolution of generations, the fittest parameters are determined to be the target plastic constitutive parameters for a specific load-displacement curve. Finally, the reverse analysis method is validated by the good agreement between the nanoindentation load-displacement curves of the experimental results and the finite element method (FEM).