Machine learning-based design exploration of clamshell telescope enclosure structure

This study presents an innovative procedure using generative design in combination with machine learning. A new workflow is developed to approximate the responses of individual model evaluations and combine them in a unified environment from which design solutions are searched. This ensures that all design considerations, constraints and objectives are taken into account simultaneously. The proposed workflow is validated on the design of the enclosure for the New Robotic Telescope (NRT), the world's largest robotic, fully autonomous, optical telescope of the four-metre-class. The proposed design for the enclosure is a curved clamshell structure with a 19-metre internal floor diameter, consisting of six segments, three on each side. The results of the study have provided insights into the behaviour of the structure and made it possible to propose final solutions that show significant improvements over the concept design in terms of total mass and operating forces.