传统文化算法的知识对于进化过程的影响是统一进行的,而知识的趋同性导致算法易早熟收敛于局部最优解.为此,提出一种新的由当前种群最优个体及其所在区域,以及当前个体共同确定的对偶知识.当对偶知识指导个体进化时,不同个体的进化方向由相对应的对偶知识所确定.对复杂函数进行了测试,所得数据表明该算法有良好的全局收敛能力及解决高维优化问题的能力.
The influence of knowledge on the evolution process in traditional cultural al- gorithms is unified. Evolving to the same direction may lead to premature convergence. A new knowledge named dual knowledge determined by situational knowledge, normative knowledge and the current individual is proposed. When dual knowledge conducts individual evolution, the direction of different individual is decided by the individual dual knowledge. Simulation of complicated functions is performed. The results indicate that this algorithm has abilities of global convergence and good performance in solving highdimensional optimization problems.