摘要
针对传统遗传算法的缺陷,提出一种基于基因位置分布差异而进行演化的改进算法。该算法利用2进制编码位置的差异性,在评价机制的基础上对种群进行动态的划分,并针对各个不同子种群的特点,使用动态的演化参数进行独立的演化操作,使得算法种群的构成类型能够保持多样性的发展,有效地抑制了"早熟"现象的发生。通过后续多个测试函数的对比实验结果表明,该算法在收敛速度、精度及稳定性上有所提高。该算法简单、易于实现、具有较强的通用性,是一种有效解决优化问题的方法。
In view of the flaws of traditional genetic algorithm, a optimized algorithm based on the diversity of gene location dis- tribution is brought out. By applying the discrepancy of binary coders location, this algorithm made a dynamic dividing to the population on the basis of evaluation mechanism and focusing on the various characteristics of each separate sub-populations, per- form independent evolution by using dynamic evolution parameters, which can ensure the developing diversity of constitution types to the algorithm population, so that the phenomena of ＂precocity＂ can be effectively restrained. Through the following comparative experiment of test functions, it shows the improvements in convergence rate, accuracy and stability： In addition, this algorithm is simpler and more achievable with stronger generality to presented as a new method to effectively solve the opti- mization problems
出处
《计算机工程与设计》
CSCD
北大核心
2012年第7期2809-2814,共6页
Computer Engineering and Design
基金
湖南省自然科学基金项目(06JJ50107)
湖南省教育厅重点基金项目(10A074)
关键词
遗传算法
非均匀演化
2进制编码
动态参数
适应度
genetic algorithm
non-uniform evolution
binary encoding
dynamic parameters
fitness
