Properly Pareto Optimality Based Multiobjective Evolutionary Algorithm for Constrained Optimization

Constrained optimization problem (COP) is converted into bi-objective optimization problem (BOP) and solved with a simple differential evolution (DE) algorithm guided by ε-properly Pareto optimality. ε-properly Pareto optimality is stronger than Pareto optimality, which results in Pareto optimal solutions at special region. Since one of the extreme point on Pareto front of BOP is the optimal solution of COP, ε-properly Pareto optimality can be used to guide generic multi-objective evolutionary algorithm to converge to the specific point. Therefore, a simple biased non-dominated sorting selection is designed to match the preference, and then it is used in DE to solve COP. Numerical experiments for several standard test functions with different characteristic illustrate that the new proposed algorithm is effective and efficient.