Hybrid fuzzy-genetic system for optimising cabled-truss structures

Abstract

This paper demonstrates an application of a hybrid fuzzy-genetic system in the optimisation of lightweight cabled-truss structures. These structures are described as a system of cables and triangular bar formations jointed at their ends by hinged connections to form a rigid framework. The optimised lightweight structure is determined through a stochastic discrete topology and sizing optimisation procedure that uses ground structure approach, nonlinear finite element analysis, genetic algorithm, and fuzzy logic. The latter is used to include expertise into the evolutionary search with the aim of filtering individuals with low survival possibility, thereby decreasing the total number of evaluations. This is desired because cables, which are inherently nonlinear elements, demand the use of iterative procedures for computing the structural response. Such procedures are computationally costly since the stiffness matrix is evaluated in each iteration until the structure is in equilibrium. Initially, the proposed system is applied to truss benchmarks. Next, the use of cables is investigated and the system’s performance is compared against genetic algorithms. The results indicate that the hybrid system considerably decreased the number of evaluations over genetic algorithms. Also, cabled-trusses showed a significant improvement in structural mass minimisation when compared with trusses.