基于克隆选择和邻域搜索的改进阴性选择算法

摘要

Based on the study of genetic RVNS algorithm(GA-RVNS), an improved RVNS based on clonal selection and neighborhood search(CSNS-RVNS) was proposed.To solve GARVNS's premature convergence and its detector sets were not optimal in N-dimensional space, a clonal selection of the immune mechanism was introduced to implement global search.A quasi-random sequence was generated to serve as the search space of clonal selection, then the optimal detector sets in sequence were obtained by clonal selection and mutation operators.The Gaussian mutation operator was proposed to get the global optimal detection sets of N-dimensional space through neighborhood search.The resulting detector sets achieve a good coverage of non-self space, and also significantly reduce the number of detector sets, thus overcome the limitations of original negative selection algorithm.Finally, experiments verify the effectiveness of the algorithm.%在研究基于拟随机序列和遗传变异搜索的阴性选择算法的基础上,针对其早熟收敛以及生成的检测器集不是N维空间最优的问题,引入了免疫机制中的克隆选择来实现检测器集的全局搜索.将生成的拟随机序列作为克隆选择搜索空间,通过克隆和变异选择操作获得空间中的优化检测器集,然后对该检测器集引入高斯变异算子,通过邻域搜索获得整个N维空间里的优化检测器集.该检测器集能很好地覆盖非我空间,而且检测器集数量相对于普通阴性选择算法也大幅减少,克服了普通阴性选择算法的局限性.最后,通过实例应用验证了算法的有效性.