Method and system for continuous multi-objective optimization of designs

摘要

According to the invention, a method for multi-objective optimization of a design comprises the steps:
providing an initial population;
selecting some evenly distributed reference points from the population;
generating randomly, for each reference point, a neighbourhood from the population,
building a probability model based on points from the neighbourhood;
generating a next population by sampling points from the probability model characterized in that
the size of the neighbourhood is adaptively adjusted.