论文部分内容阅读
提出一种相似性算子,并根据该算子建立超级个体之间的关系网,运用社团划分算法将该关系网划分成小生境;提出收敛算子,运用该算子判断种群的进化程度,以决定是否进行生境划分.根据生境中维持的共有模式,采用多位变异策略,有效维护生境中个体的多样性,避免早熟收敛.从理论上证明了该算法的收敛性.实验结果表明,该方法具有良好的收敛速度和全局搜索能力,其综合性能优于确定性排挤遗传算法,有利于并行实现.
A similarity operator is proposed and a network of super individuals is established based on the operator. The social network is divided into niches by using a community division algorithm. A convergence operator is proposed to evaluate the evolution of the population by using the operator. To decide whether to divide the habitat or not.According to the common pattern maintained in the habitat, a multi-variant strategy is adopted to effectively maintain the diversity of individuals in the habitat and avoid premature convergence.The theorem shows the convergence of the algorithm.The experimental results show that the The method has good convergence speed and global search ability, and its comprehensive performance is better than deterministic crowding genetic algorithm, which is good for parallel implementation.