Abstract:

Owing to Pareto solutions'  discrete distribution, not a continuous curve/surface, derived from a multiobjective evolutionary algorithm, the decisionmaker couldnot get Pareto solutions for totally meeting design requirements in most cases. According to the relationship between multiobjective optimization and decisionmaking machine, the preference requirement and preference function are proposed to measure the satisfaction level for objective values. And a method of Pareto improved solutions based on sensitivity analysis is used to ascertain whether there exist some more solutions which might even more meet preference requirements. Simulation results show that Pareto solutions from the candidate set are improved effectively while assisting in providing more Pareto solutions for meeting preference requirements to be the final scheme.