Optimization of 3D printed bioinspired helicoidal composites using Gaussian process regression

Abstract This work focuses on optimization of 3D printed bioinspired helicoidal composites in design space of helicoidal angle and brick aspect ratio under dynamic three‐point bending loading by performing gaussian process regression‐based Bayesian optimization method. Firstly, the impact responses and crack modes of helicoidal structures under dynamic three‐point bending loading were studied through experiments and simulations. Then, the helicoidal composites with better dynamic mechanical properties were obtained by iterations of Bayesian optimization based on gaussian process regression. Finally, the crack propagation modes of the optimized helicoidal structures under dynamic loading were investigated by analyzing the interlayer stress of the structure. The results show that the interlaminar shear stress of the optimized model is large, which is more prone to produce mixed mode I + II cracks. The mixed mode cracks cause crack deflection, which improving the fracture toughness of the structure. Highlights Bioinspired 3D printed helicoidal composites are optimized by machine learning. The helicoidal composites with better dynamic mechanical properties are obtained. The crack propagation modes of the optimized helicoidal structures are investigated. The mixed mode I + II cracks cause crack deflection.