A New Fitness Function With Two Rankings for Evolutionary Constrained Multiobjective Optimization

Among the constraint-handling techniques (CHTs) in constrained multiobjective optimization, constrained dominance principle (CDP) is simple, flexible, nonparametric, and easy to be embedded into multiobjective evolutionary algorithms. However, CDP always prefers constraints to objectives, which tends to cause premature convergence. To overcome this drawback, we propose a new CHT on the basis of CDP. In our CHT, the fitness function of each solution is defined as the weighted sum of two rankings: one is the solution's ranking based on CDP and the other is the solution's ranking based on Pareto dominance (i.e., without considering any constraints). It is evident that these two rankings favor the feasibility and optimality of each solution, respectively. More importantly, the weight employed in the fitness function is related to the proportion of feasible solutions in the current population, enabling the population to adaptively balance constraints and objectives in the evolutionary process. The effectiveness of our CHT is evaluated on three test suites and the experimental results demonstrate that our CHT shows better or highly competitive performance compared with other representative CHTs.

Automated summary

We propose a new constraint-handling technique based on a weighted sum of rankings using constrained dominance principle and Pareto dominance, which adaptively balances constraints and objectives in evolutionary optimization process. Experimental results show that our technique outperforms other representative constraint-handling techniques.