| About this Abstract |
| Meeting |
2024 Annual International Solid Freeform Fabrication Symposium (SFF Symp 2024)
|
| Symposium
|
2024 Annual International Solid Freeform Fabrication Symposium (SFF Symp 2024)
|
| Presentation Title |
Automated Design of Optimal Plate Lattice Structures Enabled by Machine Learning |
| Author(s) |
Charles Wade, Robert MacCurdy |
| On-Site Speaker (Planned) |
Charles Wade |
| Abstract Scope |
Plate lattice structures offer a promising solution for impact absorption due to their tunable geometric parameters which can be optimized to improve energy absorption across a range of impact energies. This talk covers our work on automated design synthesis and surrogate modeling for optimizing these structures. We use a multi-objective heuristic optimization process with FEA simulations to discover Pareto-optimal designs among thousands of candidates. To accelerate this process, we train a neural network to predict impact absorption profiles, significantly speeding up the discovery of optimal lattice geometries. By using a smaller set of simulations for training, the network accurately predicts performance across diverse geometries and impact scenarios. This surrogate model, combined with the optimizer, expands the search space to millions of design candidates. We benchmark our plate lattices against established impact absorbing geometries, to demonstrate how automated design synthesis and surrogate modeling enable efficient optimization, offering superior tunability and performance. |
| Proceedings Inclusion? |
Definite: Post-meeting proceedings |