A view on automated neural graph topology generation and a viable direction of innovation

Abstract Compliant mechanisms are a class of mechanisms that achieve desired force and motion transmission tasks by undergoing elastic deformations as opposed to rigid-body displacements in the conventional rigid-link mechanisms. Most of the previously reported synthesis studies in compliant mechanisms related to either partially-compliant mechanisms or fully-compliant mechanisms with joint compliance. Methods developed for fully-compliant mechanisms with link compliance addressed the issue of topology generation for desired deflections at discrete points on the mechanism. This paper presents a new, first-principles based synthesis procedure for fully-compliant mechanisms with link compliance — that is, distributed-compliant mechanisms — for continuous shape change requirements in a particular segment of a mechanism. The general approach presented in this paper for the synthesis of distributed compliant mechanisms is shown to be well suited for application in the design of adaptive structures, an emerging class of high-performance structural systems. The current trend in the design of adaptive structures is to embed structures with force or strain inducing “smart” materials to serve as distributed actuators. Potential advantages of using the distributed compliance scheme over the distributed actuation scheme in the design of adaptive structures include a significant reduction in the number of required actuators and controls.

Download Full-text

An Improved Shape Annealing Method for Truss Topology Generation

6th International Conference on Design Theory and Methodology ◽

10.1115/detc1994-0033 ◽

1994 ◽

Author(s):

Giridhar Reddy ◽

Jonathan Cagan

Keyword(s):

Simulated Annealing ◽

Optimization Method ◽

Geometric Constraints ◽

Truss Structures ◽

Model Design ◽

Topology Generation ◽

Truss Topology ◽

Gradient Based ◽

Annealing Algorithm ◽

Annealing Method

Abstract A method for the design of truss structures which encourages lateral exploration, pushes away from violated spaces, models design intentions, and produces solutions with a wide variety of characteristics is introduced. An improved shape annealing algorithm for truss topology generation and optimization, based on the techniques of shape grammars and simulated annealing, implements the method. The algorithm features a shape grammar to model design intentions, an ability to incorporate geometric constraints to avoid obstacles, and a shape optimization method using only simulated annealing with more consistent convergence characteristics; no traditional gradient-based techniques are employed. The improved algorithm is illustrated on various structural examples generating a variety of solutions based on a simple grammar.

Download Full-text