Study on hardness prediction and parameter optimization for carburizing and quenching: an approach based on FEM, ANN and GA

Abstract A proper hardening depth is critical to the load-bearing capacity of a part, and heat treatment, including carburizing quenching, can highly determine hardness distribution in part’s surface after manufacturing. This paper proposes ‘hardness prediction parameter optimization’ approach that deploys finite element method (FEM), artificial neural network (ANN), Genetic Algorithm (GA), describe relationships between carburizing/quenching parameters depths conversely optimized for given depth. First, numerical models carburizing, field are built respectively. And based on these models, simulation model designed predict carbon content, microstructure part. BP then trained by using data obtained from simulation, part established. The optimization finally established through GA, which effectiveness ‘prediction-optimization’ verified series experiments. predicted proposed holds 10% relative error measured quenching experiment. treatment process work as meaningful reference treatment.