DARTS-NETGAB: Design Automation and Real-Time Simulation Using Neural Networks Ensembles for Turbocompressors on Gas-Bearings
Published:
Link: http://dx.doi.org/10.1115/detc2024-143135
Overview
DARTS-NetGAB combines ensemble neural-network models, an interactive Panel/Bokeh interface, and parametric CAD generation for gas-bearing turbocompressors. Users can modify design parameters, inspect predicted performance maps, and generate corresponding three-dimensional geometry within the same workflow. The conference paper reports the model architecture, validation results, interface, and connection to the ParaturboCAD geometry pipeline.