Process optimization of selective laser melting 316L stainless steel by a data-driven nonlinear system

Selective laser melting (SLM) provides a great degree of design freedom, but because the forming process is accompanied by multi-scale and multi-physics phenomena, the computational cost of modelling the forming process remains quite high. The cost of testing and simulation can be reduced by creating a model based on a nonlinear system, fitting the nonlinear function between input and output, and choosing the best option for forming quality. In this study, a nonlinear system was created by integrating data mining and statistical inference technologies, and a mix of simulation and experiment was utilized to create a specimen with the desired properties. Firstly, the density of 316L stainless steel specimens formed by SLM with various layer thicknesses was obtained, the important parameters impacting density were examined, and the effect of layer thickness on specimen density was reported. Furthermore, the nonlinear neural network optimization system was built through training and testing in order for the trained network to forecast the nonlinear function’s output. Finally, the system is utilized to forecast density in high-efficiency moulding mode, and the test results match the anticipated data well.