为提高人工免疫算法求解旅行商(TSP)问题的效率,设计了一种疫苗的动态提取策略.该策略通过对记忆种群的一个随机子集进行线性复杂度的集合求交集运算,不仅能自适应地提取单个基因疫苗,而且能获得长度大于1的多基因疫苗组.随着迭代的进行疫苗长度的自适应增加,降低了原TSP问题的规模,压缩了算法的搜索空间.与其他疫苗策略相比,该策略无需人为参与,能更准确地预测最优路径中的边,帮助算法获得更高质量的解.
To enhance the efficiency of artificial immune algorithms for Traveling Salesman Problem (TSP), we designed a dynamic vaccination strategy. The proposed vaccination obtains both single- vaccines and multi-vaccines by applying a linear complex intersecting operation on a random subset of the memory cell. The lengths of vaccines increase with iteration, which depresses the problem size and algorithm's searching space. Compared with other vaccination strategies, the proposed strategy is unsupervised, which makes more accurate prediction of edges in the best tour and helps immune algorithms to maintain better solution paths.