Enhancing topology optimization for multi-objective using sheet-based TPMS and CFRP: an ANN and NSGA-II approach

Purpose This study aims to enhance a decision-support system that offers optimum solutions to obtain the optimal topology using a sheet-based triply periodic minimal surface (TPMS) and carbon fiber-reinforced polymer (CFRP). Design/methodology/approach Within the scope of this study, the mechanical responses resulted by changing the orientation, type, size, relative density and graded relative density of the unit cells likewise the number of composite layers were examined. Wrapped around the sheet-based TPMSs, the CFRP allowed the lattices to absorb more energy during deformation, maintaining their shape integrity. An artificial neural network (ANN) was trained to reveal the relationships between the design parameters and mechanical properties. Findings According to SHAP values, the highest significance in the ANN model was determined as mass, graded relative density, cell size and number of composite layers. The significance of mass was greater than the sum of the importance of the other design parameters. An approximately linear relationship existed between the design parameters and peak crushing force, mean crushing force, energy absorption and plateau stresses, whereas specific energy absorption (SEA) and crushing load efficiency (CLE) had a complex relationship. Originality/value The non-dominated sorting genetic algorithm-II (NSGA-II) was used to find the optimum solution from the complex relationship between the design parameters and SEA and CLE. Design parameters for optimum crashworthiness were determined using NSGA-II, a heuristic optimization method using an ANN as the objective function.