Conceptual design of structures using machine learning
We developed and validated a computer-assisted tool based on Non-dominated Sorted Genetic Algorithm II (NSGA-II) that can be used to analyse a wide range of safe, economical and ecological options for the conceptual design of buildings.
The design space starts from a design brief (with only a little information about the site characteristics and project objectives). The solutions are explored with the material, grid size, floor type, lateral resistance, and foundation system variables. In a short computational time (< 2 min per run), users are provided with a Pareto graph of a large set of feasible solutions (in terms of cost, embodied CO2 emissions and free space) that an engineer would not be typically able to evaluate within a traditional conceptual design process.
More details can be read in the following articles:
Image classification using conventional neural networks:
We are developing a convolutional neural network tool that is capable of assessing the images for several criteria i.e. corrosion, damage, connection type. The objective of the tool is to support the engineers and architects while giving re-use/recycle decisions.
Research is underway. Soon they will be published.