An Evolutionary Clustering Approach to Pareto Solutions in Multiobjective Optimization

Pareto solutions in multiobjective optimization are very problematic to measuring the characteristics of solutions for engineering design because of their considerable variety in function space and parameter space. To overcome these situations, a clustering-based interpretation process for Pareto solutions is considered. For better competitive clustering algorithm, we propose an evolutionary clustering algorithm — ECA. The ECA requires less computational effort, and overcomes local optimum of the K-means clustering algorithm and its related algorithms. Effectiveness of the method is examined in detail through the comparison with other algorithms.