Incorporation of Fuzzy Preferences into Evolutionary Multiobjective Optimization

A method for incorporating fuzzy preferences into evolutionary multiobjective optimization is proposed. After introducing three commonly used models for describing fuzzy preferences, a method to convert fuzzy preferences into realvalued weight intervals is suggested. It is argued that to convert fuzzy preferences into interval-based weights is more consistent with the motivation of using fuzzy preferences than to convert them into single-valued crisp weights. The weight intervals are combined with the evolutionary dynamic weighted aggregation to obtain the preferred Pareto-optimal solutions. Simulation examples are given to show how the desired Pareto-optimal solutions can be obtained