Structural system design is the process of giving form to a set of interconnected components\nsubjected to loads and design constraints while navigating a complex design space. While safe\ndesigns are relatively easy to develop, optimal designs are not. Modern computational optimization\napproaches employ population based metaheuristic algorithms to overcome challenges with the\nsystem design optimization landscape. However, the choice of the initial population, or ground\nstructure, can have an outsized impact on the resulting optimization. This paper presents a new\nmethod of generating such ground structures, using a combination of topology optimization (TO) and\na novel system extraction algorithm. Since TO generates monolithic structures, rather than systems,\nits use for structural system design and optimization has been limited. In this paper, truss systems\nare extracted from topologies through morphological analysis and artificial intelligence techniques.\nThis algorithm, and its assessment, constitutes the key contribution of this paper. The structural\nsystems obtained are compared with ground truth solutions to evaluate the performance of the\nalgorithms. The generated structures are also compared against benchmark designs from the literature.\nThe results indicate that the presented truss generation algorithm produces structures comparable\nto those generated through metaheuristic optimization, while mitigating the need for assumptions\nabout initial ground structures.
Loading....