作　　者： ; ; ; ;

机构地区： 华南农业大学

出　　处： 《计算机仿真》 2013年第2期316-321,共6页

摘　　要： 针对标准的量子遗传算法(QGA)应用于数值优化时容易早熟收敛而陷入局部最优的问题,引入k位变异子空间概念对Q-bit变异概率分布进行了分析,传统随机变异机制和QGA自蕴变异机制存在冲突。为此提出一种用观测状态的阶段式大尺度变异机制(SLVMBOO),并将SLVMBOO变异算子嵌入到量子旋转策略表中,实现起来简单高效。通过典型复杂函数测试表明SLVMBOO使得QGA应用于数值优化时能有效地避免早熟收敛、跳出局部最优,而且全局寻优能力优于其它方法。 Standard quantum genetic algorithm （QGA） applied to numerical optimization is easy to converge to local optima because of premature. To solve this problem, this paper analyzed the mutation probability distribution of Q - bit by introducing the k - bit variation subspaee conception, and pointed out the conflict of traditional random mutation mechanism and the QGA self - implied variation mechanism. Based on these analyses a novel Stage Large - scale Variation Mechanism Based on Observation （SLVMBOO） was proposed. Mutation operator of SLVMBOO em- bedded in the quantum rotation policy table is simple to implement and it is highly efficient. Typical complex function test showed that SLVMBOO makes the QGA effectively avoid the premature convergence and successfully jump out of local optima, when applied to numerical optimization. QGA -SLVMBOO＇s global optimization ability is superior to other methods in the literature.

关 键 词： 量子计算 量子遗传算法 变异机制 变异概率分布 数值优化

领　　域： [自动化与计算机技术] [自动化与计算机技术]