A memory-based technique for optimal structural design

Abstract

Tabu search is a heuristic optimisation technique that draws on ideas in the field of artificial intelligence. It is an iterative improvement algorithm characterised by a semi-intelligent approach to the search of the solution space. A feature of the algorithm is the use of primary and secondary memory facilities to avoid entrapment by local optima and to diversify the search for a global optimum. Tabu search is a relatively new technique that has recently been applied successfully to problems in manufacturing; for example, scheduling and facilities layout. In this paper it is applied to the weight minimisation of a space truss subject to applied loads. This optimal structural design problem is chosen as a vehicle for tabu search because it is challenging in the sense that there are numerous local minima. The design variables are the cross-sectional areas of the truss members, which may be selected from a finite set of discrete values. Hence, in mathematical terms, the minimisation problem is combinatorial in nature. An implementation of tabu search in a structural design context is presented which features λ-depth neighbourhoods and a search back-track facility. Furthermore, through the use of reliability constraints, the presented formulation shows how the effects of stochastic variations in material strength and applied loading may be analysed.