Abstract Scope |
Lattice structure, an important structural design feature in Additive Manufacturing (AM), is commonly used for lightweighting and multiple functions such as thermal dissipation, vibration damping and energy absorption. In this work, a Machine Learning (ML)-based inverse generator, capable of identifying optimal 3D lattice unit cells with desirable mechanical properties, is trained using the dataset computed using homogenisation method. The proposed inverse generator can be applied to multi-scale structural design where grading of lattice is needed. As an example, designs from the proposed methods are generated for lattice core sandwich structures. Within the core design domain, the spatially-varied lattice parameters are identified through the inverse generator. The benchmark example considered proves that the proposed method is capable of generating structures that are high performing (e.g. stiffness, energy absorption and impact resistance) at minimal computational cost. |