提出了一种多目标进化算法中解集分布度逐步评价方法.定义了一种基于角度的坐标,避免了算法因收敛性不同对分布性评价的影响;利用了解集均匀分布具有的对称性,把整个目标空间从大到小划分成不同的对称区域,逐步进行分布度评价.实验结果表明,该方法能精确的评价解集的分布情况.
A measurement for stepwise diversity evaluation of solutions in multi-objective evolutionary algorithm was proposed.In this method, a coordinate based on angle was defined, which can avoid the influence caused by different extent of convergence in diversity evaluation. Furthermore, considering the symmetry of uniformly distributed solutions, the whole objective space was divided and subdivided into different symmetric regions for evaluation. Experiments show that the method can evaluate diversity of solutions accurately.