Quantitative Relationship Analysis Between Mechanical Properties and Microstructures of Al-7Si Aluminum Alloys by Artificial Neural Network

An artificial neural network model with high accuracy and good generation ability was developed to predict and optimize the mechanical properties of Al-7Si alloys. The results show that Al-7Si alloys with tensile strength of 310~350 MPa, elongation of 3%~12%, and different microstructures are obtained by controlling the holding pressure (85~300 kPa) and cooling rate (1~10 k/s) of the casting process. The quantitative correlation relationships of the mechanical properties with microstructures of the secondary dendrite arm spacing (18.56~33.04 μm), area of eutectic Si phase (6.37~13.37 μm²), area fraction of porosity defects (0%~0.363%), and area fraction of maximum Fe-rich intermetallics (0%~0.06%) in the alloy were established. The individual and combined influences of these microstructure characteristics on the mechanical properties were simulated. Both tensile strength and elongation are inversely related to the above-mentioned structural characteristics, and the presence of defects and Fe-rich intermetallics have great adverse effects on the properties of the alloy. Therefore, narrowing the dendrite spacing (<20 μm), modifying the eutectic Si phase (<12 μm²), and controlling the porosity defects (<0.35%) and the morphology of the Fe-rich intermetallics are keys to prepare high-performance aluminum alloys.