Functionally graded cellular structures usually exhibit excellent energy absorption properties, showing great promise to potentially replace conventional uniform cellular structures for next-generation passive protection systems. Meanwhile, Additive Manufacturing (AM) technologies have enabled the rapid high-resolution prototyping of functionally graded gyroid-based cellular structures. Due to the complexity of the simulation for functionally graded gyroid-based cellular structures, developing an efficient computational method with minimal computational cost to find the optimal design for energy absorption becomes a significant challenge. This paper presents a physics-informed machine learning method for the design of functionally graded gyroid-based cellular structures that can be additively manufactured to achieve an optimized energy absorption performance. We first generate simulation results of different functionally graded gyroid-based cellular structures. Then we use these results as the training set for our machine learning surrogate model. Finally, we adopt the adaptive sampling method to improve the model. This proposed method provides a less computationally expensive approach to finding the optimal design for functionally graded gyroid-based cellular structures, which can achieve the maximum energy absorption at a specific average volume fraction for passive protection systems, considering manufacturing restrictions. Using the proposed method with minimal computational cost, the optimal design of the passive protection system can be obtained, and the optimized functionally graded gyroid-based cellular structures are additively manufactured and tested for the design validation studies.

• Published and presented 1 paper as the 1st author on IISE 2022