基于网格和Hamming距离的伪并行遗传算法

为了克服简单遗传算法“早熟”现象及慢速收敛性,提出基于网格和Hamming距离的伪并行遗传算法,即用随机选取的决策变量把整个搜索空间网格划分,在每个网格内用基于Hamming距离的随机化方法初始化子种群,同时引入多遗传算予和无约束移民策略和替换策略。该算法在一定程度上克服了引发算法早熟的“超级个体”和“封闭竞争”问题。测试实例说明,相比其他遗传算法,该算法更有效地抑制了“早熟”现象,其运行速度和精度也得到了相应的提高。

Pseudo-parallel Genetic Algorithms Based on Grid and Hamming-distance

For overcoming the premature and low convergence of simple genetic algorithms (SGA), this paper puts forward pseudo-papallel genetic algorithm based on grid and Hamming-distance (PPGAGH), that is partition the whole search space using grid, and generates initial subpopulation in each grid using random based on Hamming-distance. Multi-genetic operators and unrestricted migration strategy and substitution strategy arc also introduced. This algorithm overcomes "super-individual" and "close competition" problems which lead premature of SGA in a way. Test results show that PPGAGH can not only restrain the premature more efficiently but also improve convergence speed and accuracy compared with other GA.