A large class of engineering design situations can be described by combinations of discrete variables taken from finite sets. In these circumstances design optimization can be focussed on a simple combinatorial search of a finite design "space", though this is usually very large in size.

The genetic algorithm (GA) provides a basis for such a search and considerable success has been achieved in practical applications (Goldberg et al 1987, 89, Jenkins 1991, 92, 93, Rajeev & Krishnamoorthy 1992).

The space condensation heuristic to be described was developed as an enhancement to the GA in the context of structural design optimization where it was found that preferred values of discrete variables were frequently indicated by the algorithm during the processing.

The heuristic uses a structured record of the parameter values selected by the algorithm. Those associated with "good" solutions are recorded positively and those associated with "poor" solutions are recorded negatively. The record is then used to progressively reduce the size of the space being searched.

The success of the heuristic with genetically driven search suggests that it may be useful with other combinatorial search algorithms.

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description
purchase this book (price £60 +P&P)