A Tensegrity Based Structure Optimization Framework
Loading...
Date
2014-08-25
Authors
Jdanov, Vyacheslav
Keyword
Tensegrity , Unsupervised , Framework , Optimization
Abstract
This thesis presents a tensegrity based structure optimization framework. A tensegrity structure consists of rods (compressed elements) and elastics (tensile elements) connected into a stable structure. A tensegrity structure diffuses the forces applied to any one of its elements to the entire network of connected elements, making the whole of the structure flexible and strong.
This thesis focuses on structural optimization problems that are bound in the physical world and subject to real-world laws; examples include optimization of a bridge design and optimization of an electrical grid. The framework presented in this thesis exploits the properties of tensegrity networks as an efficient, powerful, low-level heuristic to steer the optimization.
The tensegrity modeling framework stipulates that a tensegrity model consists
of three components: an optimization goal (specifies the method by which the performance of an instantiation of the tensegrity model is evaluated), a graph specification (defines the properties of a tensegrity graph, including edge labels and attributes that can appear in this graph), and an adaptation rule set (specifies the change over time in the attributes of a tensegrity graph).
A tensegrity graph provides a versatile platform for structural modeling and optimization, because attributed nodes and edges can be used to satisfactorily represent the essential properties of a large variety of optimization problems. This is illustrated in the three examples: a detailed and instantiated model of fascia, a conceptual bridge model, and a model abstracting the intricacies of an electrical grid. The components of each model are very easy to modify, allowing for rapid prototyping and insight into the significance of the variables altered between prototypes. The framework is generic, allowing for models of varying degrees of abstraction while remaining intuitive and applicable to the structural optimization problems they represent.