Process parameter optimisation for selective laser melting of AlSi10Mg-316L multi-materials using machine learning method

Abstract

The present work focuses on process parameters optimisation for selective laser melting (SLM) of AlSi10Mg-316L multi-materials using machine learning method. The mechanical properties of the multi-material samples were measured at different process parameters. These process parameters and properties data were used to train and validate the machine learning model. A multi-output Gaussian process regression (MO-GPR) model was developed to directly predict the multidimensional output to overcome the limitations of the standard Gaussian process regression (GPR) model. Based on the prediction data, process parameter maps were constructed, and the optimal process parameters for different compositions were selected from the process parameter maps. The results showed that the laser power, scan velocity and hatching space have an important influence on the density and surface roughness of the samples. Results also indicated that there is no linear functional relationship between the optimal volumetric energy density (VED) values and the AlSi10Mg-316L compositions.

异质金属激光增材制造研究及应用进展（特邀）

“…(a) Schematic diagrams of the four types of joining strategies [77] ; (b) LPBF direct joining of Cu10Sn/Ti6Al4V and its interface [19] ; (c) fingercrossed 316L/CuSn10 interfacial structure and samples formed by SLM [81] ; (d) bionic SS316L/IN625 interface structure formed by LDED [82] ; (e) Ti6Al4V/SS316 stainless steel heterometal connection through the intermediate layer Cu10Sn using SLM [79] ; (f) inconel 718/Ti64 heterometal connection through VC -Inconel 718 -Ti64 combination bonding layer using LENS [80] ; (g) martensitic stainless steel/austenitic stainless steel connection through gradient layer using LENS [78] 1002304 -11 9 Influence of process parameters on the microstructure and properties of heterogeneous metal components. (a) Variation of the relative density of LMDformed GH4169/K417G heterogeneous metal components with laser power [76] ; (b) LMD parameter optimization strategy and IN625/304L heterogeneous metal components prepared under the optimal parameters [90] ; (c) Ti6Al4V/ AlSi10Mg heterogeneous metal interfaces formed by LPBF under different process parameters [30] ; (d) NiTi/CuSn10 heterogeneous metal interfaces formed by SLM under different process parameters [12] 表 6 经典异质材料体系的最优/适配工艺参数 Table 6 Optimization/processing parameters for classical heterogeneous material diagram of the integrated highspeed imaging device of the LMD equipment and highspeed imaging pictures of the Cu deposition process [99] ; (c) LDED system with integrated monitoring and sensing device [100] ; (d) (e) schematic diagram of the insitu Xray imaging device integrated into the LPBF system and the photos of the Inconel 718/316L forming process captured under different processing parameters [101] ; (f) schematic diagram of integrated optical device for LDED systems [102] 亮点文章•特邀综述 adaptive mesh and boundary conditions for thermal analysis [103] ; (b) machine learning flowchart for SLM parameter optimization [104] ; (c) simulation of powder bed temperature distribution with different gradient IN718 components during LPBF manufacturing of IN718/Ti6Al4V [105] ; (d) simulation of deformation and stress distribution in LDED manufacturing of Cu/ SS304L [83] 表 7 异质金属激光增材制造的监测技术 Table 7 Monitoring techniques for laser additive manufacturing of heterogeneous metals with or without preheating [110] ; (d) (e) comparison of microstructures of CuBe/H13 formed by LMD with or without preheating…”