A self-organizing map approach for constrained multi-objective optimization problems

Abstract There exist many multi-objective optimization problems (MOPs) containing several inequality and equality constraints in practical applications, which are known as CMOPs. CMOPs pose great challenges for existing evolutionary algorithms (MOEAs) since the difficulty balancing objective minimization constraint satisfaction. Without loss of generality, distribution Pareto set a continuous m-objective CMOP can be regarded piecewise manifold dimension ( m − 1). According to this property, self-organizing map (SOM) approach constrained is proposed article. In approach, we adopt strategy two population evolution, one evolved by considering all other used assist exploring areas. stage, each assigned discovering structure decision space. After topological mapping, utilize extracted neighborhood relationship information generate promising offspring solutions. Afterwards, neuron weight vectors SOM updated surviving offsprings. Through make efficiently converge feasible region with suitable levels diversity. experiments, compare method state-of-the-art approaches using 48 benchmark problems. The evaluation results indicate that overwhelmingly superior performance over peer on most tested source code available at https://github.com/hccccc92918/CMOSMA .