Data-Driven Design and Performance Optimization of Hybrid Micro-Lattice Architectures for Lightweight Applications

Lattice structures afford lightweight designs that can vary in terms of stiffness, strength, energy dissipation capacity, and surface area to volume ratio; nevertheless, traditional lattice structures of the same topology represent limited options for meeting the conflicting biomechanical objectives at once. This work describes a computational design and screening strategy for hybrid micro-lattices made up of BCC, octet-truss, diamond, cubic, FCC, and cuboctahedral unit cells. Predictions made from computational results demonstrated that the relative-density range would be widened from 3–25% for hybrid structures compared to 2–5% for single architectures, resulting in a more than fivefold increase (p < 0.05). Significant associations were found between the compressive stress and specific strength (r = 0.91) and between stiffness and specific stiffness (r = 0.96; p < 0.01), while the included biocompatibility factor showed no significant association with mechanical properties (r < 0.10; p > 0.05). BCC possessed the predicted compressive strength value of 155 MPa, whereas cuboctahedral lattices had the predicted elastic modulus value of 7.2 GPa. From hybrid structures, Octet-BCCZ and Octet-FCCZ showed good predicted strength/weight properties, but Octet-Diamond-Octahedron possessed the best predicted biomechanics balance, including 0.21 mm of deformation, 138 MPa of compressive strength, and 6.1 GPa of elastic modulus. XGBoost algorithm gave the best prediction accuracy (R 2 = 0.560; RMSE = 0.090). Optimization by the Pareto method helped to identify the non-dominated architecture options for balanced biomechanics. All findings have been obtained via computational analysis and machine learning predictions. All reported findings have been acquired solely through computational simulation studies and machine learning-based predictions. Experimental confirmation using L-PBF fabrication, mechanical testing, and biological evaluation has not been attempted and still needs to be explored in the future in order to determine the feasibility of the proposed designs.