The application of the genetic algorithm (GA) to structural design optimization can involve large numbers of design variables and consequently long binary strings to represent individual designs. Some form of partitioning is needed to obtain effective handling of these long strings and to enable efficient processes of selection and recombination to be put in place. An intuitively attractive basis of string partitioning is variable-by-variable and this has the added advantage of easily accommodating unequal string lengths and enabling progressive reductions in partial string lengths in a combinatorial space condensation heuristic (Jenkins). The paper explores a possible approach to the assessment of these partial strings based on records of string fitnesses and parameter values selection numbers. It is shown that a partial string fitness operator can identify relatively "good" and "bad" in partial string fitnesses and adjust fitness values accordingly.

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 £55 +P&P)