This paper explores how Conditional Generative Adversarial Networks (cGANs) can be used as a tool for structural adaptation in hybrid systems such as gridshells and plate structures. The cGAN is trained to identify stress trajectories to enhance structural performance. Height-to-stress mappings act as the basis for testing how cGANs can assist in geometric adjustments decision making by providing quickly produced generalized patterns of a design system’s structural identity. By embedding structural analysis and quantitative constraints into a machine learning dataset, the method provides iterative feedback on stress distribution rather than treating computational analysis as a terminal output. There are two deployment strategies: 1. Applying various magnitudes on a plate structure to evaluate the model’s ability to distinguish between different levels of force based on limited representational differences in the cGANs training, 2. Redistributing material in response to generatively mapped stress on a shell structure.
Brooking et al. (Mon,) studied this question.