Determination of Elastoplastic Properties of 2024 Aluminum Alloy Using Deep Learning and Instrumented Nanoindentation Experiment

The instrumented nanoindentation technique has been widely used to measure the tensile properties of various materials,for its simple specimen preparation and nearly nondestructive testing processes.In this paper,a novel inverse method is established for measuring the elastoplastic properties of Al 2024 alloy.The grid indentation experiments are performed on Al 2024 material.The obtained experimental load–displacement(P–h)data exhibit obvious scatter characteristics.The artificial neural network(ANN)model with tunable hyper-parameters is adopted to establish the forward relationship between elastoplastic parameters and indentation load–displacement snapshot.An objective function for quantifying the error norm between predicted and experimental P–h snapshots is established.The parameter identification problem is solved using the“interior-point”constraint optimization algorithm..The identified material properties show good agreement with the tensile data,and the error values are−8.66%for elastic modulus,1.08%for yield stress,and 6.90%for hardening exponent.The sensitivity of numerical results to experimental uncertainty is analyzed,and the error bound of experimental data is determined.The results of sensitivity analysis indicate that the proposed inverse method in the work is very effective and reliable.