Bi-goal evolution for many-objective optimization problems