The estimation of the thermal performance of heat sinks manufactured by direct metal laser sintering based on machine learning

The study is about designing, modelling, and manufacturing the heat sinks using AlSi10Mg powder material by the Powder Bed Fusion (PBF) additive manufacturing method. The design of the heat sink consists of different with lattice structures X, Grid, and Rhombic structures. In this study, the designed heat sinks were manufactured using the Direct Metal Laser Sintering (DMLS) technique on an EOSINT M290 system. The study focused on examining the effect of the lattice structures regarding the resultant surface area and the thermal performance of the heat sinks. The results show that the lightweight lattice structure heat sinks performed close to the thermal performance of traditional heat sinks. This is due to the fact that the lattice heat sinks were so high due to lattice meshes that it negated the positive effect of the greater surface area. The X-lattice heat sink produced by the DMLS process was shown the best thermal performance by spreading the temperature over time and regularly reduce. The results showed that using three machine learning and three ensemble machine learning algorithms, the most suitable heat sink can be selected with an accuracy of 99.8% by the the ensemble-ANN algorithm. For the obtained model, an interface program was performed for the users to see and follow the experimental results and the results obtained from the model at the same time.