Controlled mutation in evolutionary structural optimization

The paper deals with minimum weight design applying Evolutionary Strategy ( ES), improved by controlled mutation. Applied selection and crossover are t ypical for ES. Mutations, however, are controlled by state variables, in th is case by stresses. After crossover, from each population, a voted number of chromosomes are inspected from the point of view of stresses occurring i n structural members (genes). Changes of cross-sectional areas are introduc ed depending on the minimum, or maximum stresses. 124 runs of the algorithm were performed on a bench-mark problem, with known optimum solution. The n umber of exact solutions, applying controlled mutation, was four times larg er than in the case of a simple ES algorithm.