基于偏好排序淘汰NSGAII算法的短波网络多区域重点覆盖优化方法

Multi-areas Outstanding Covering Optimization Method of HF Network Based on Preference Ranking Elimination NSGAII Algorithm

在采用偏好NSGAII算法求解多子区域重点覆盖的短波网络频率优化指配时,针对算法中非支配排序耗时较多的问题,该文提出一种偏好排序淘汰的NSGAII算法。在进行非支配排序前,根据解的偏好评价排序结果淘汰一部分偏好评价较差的解,减少参与非支配排序的解的数量从而减少求解时间,同时降低偏好评价结果较差的个体解被选中进行交叉、变异的概率,提高算法的求解效率和求解效果。在进行的48组数据测试中,该文算法在其中38组决策解偏好评价结果和求解时间同时最优,相同迭代次数时相比偏好NSGAII算法节省27%的求解时间。结果表明通过偏好排序淘汰机制的引入,更好利用了偏好信息,使算法用较少的时间求得更好的偏好解。

This paper proposes a preference ranking elimination NSGAII algorithm to deal with the time-consuming issue of the preference NSGAII algorithm in optimizing HF network frequency assignment in multi-areas outstanding coverage. The proposed algorithm sorts and eliminates solutions according to their preference evaluation priori to the non-dominate sorting. By eliminating solutions with low ranking, the number of solutions participates in non-dominate sorting is reduced. The calculation time and the probability of selecting low ranking individuals for crossover or mutation are both decreased. The proposed algorithm simultaneously achieves the best performance and least calculation time in 38 of 48 sets experiments. Constrained with the same iteration number, the proposed algorithm saves 27% of computation time against the preference NSGAII algorithm. Experimental results show that by adopting preference evaluation sorting, the proposed algorithm takes less time and obtains a better solution.