Breeding Topology: Special Considerations For Generative Topology Exploration Using Evolutionary Computation (2008)
article⁄Breeding Topology: Special Considerations For Generative Topology Exploration Using Evolutionary Computation (2008)
abstract⁄Topology optimization of engineering structures has long been a topic of research scrutiny. Many methods have been successfully developed for the determination of continuum structures. Some of these techniques, for example the homogenous method, have also been adapted for use with discrete structural frames or trusses. Most commonly the topology optimization of truss structures is carried out with the aid of a ground structure, a simple raster that describes potential joint locations. Although this simplifies the computation, it greatly limits the range of potential solutions that fit the gridded raster. Additionally, when using Evolutionary Computation EC methods, the level of computational intensity increases exponentially with the size of the ground structure making anything above a very modest level of complexity impractical to process. This paper demonstrates several practical techniques that can be used with EC, and more specifically Genetic Algorithms, when applied to topology exploration of discrete structures. First a method of chromosome coding that avoids the use of ground structures is shown. Then specific genetic recombination techniques are illustrated that are well suited for breeding different topologies. The combined techniques are demonstrated in a topology design problem. The paper concludes with a discussion of advantages of EC over traditional optimization methods in the area of overall form design.
|
|
| Year |
2008 |
| Authors |
Von Buelow, Peter. |
| Issue |
Silicon + Skin: Biological Processes and Computation, |
| Pages |
346-353 |
| Library link |
N/A |
| Entry filename |
breeding-topology |
