论文部分内容阅读
为提高算法NSGA-II-DE解决含有复杂Pareto解集优化问题的性能,分析原NSGA-II中拥挤度计算公式和排挤机制的缺陷,并以NSGA-II-DE算法为基本框架,将传统拥挤度排序策略改为包含有角度信息与伪半径的二维信息排序策略.在拥挤度排挤机制中加入数量级阈值的干预,提出改进算法2D-Thr.选取多样度、收敛度和分布度3个评价指标进行量化计算,并与NSGA-II-DE、原NSGA-II、MACPSO进行比较.仿真结果表明,改进算法不仅有效继承了原算法优良的收敛性,而且提高了Pareto前沿的分布度.
In order to improve the performance of the algorithm NSGA-II-DE to solve complex optimization problems with complex Pareto solutions, this paper analyzes the defects of the original NSGA-II congestion degree calculation formula and crowding out mechanism, and uses the NSGA-II-DE algorithm as the basic framework, Degree sorting strategy is changed into a two-dimensional information sorting strategy including angle information and pseudo-radius.An order-of-magnitude threshold intervention is added to the crowding-out mechanism, and an improved algorithm 2D-Thr is proposed to select three kinds of evaluation indexes, ie, degree of diversity, degree of convergence and degree of distribution And compared with NSGA-II-DE, NSGA-II and MACPSO.The simulation results show that the improved algorithm not only inherits the good convergence of the original algorithm effectively, but also improves the distribution of Pareto frontier.