论文部分内容阅读
为提高多元宇宙优化算法(MVO)的全局探索和局部开采性能,提出一种耦合横纵向个体更新策略的改进MVO算法(IMVO)。横向更新策略是建立在宇宙种群层级的一种水平迁移进化机制,通过引入加权学习因子保证子代个体同时向多个父代宇宙继承位置信息以改善种群的个体多样性和算法全局探索性能,适定性修正虫洞存在概率表达以保证种群个体间的充分信息交互;纵向更新策略是基于宇宙个体层级的一种纵向自我学习进化机制,根据最优宇宙历史信息通过模拟认知的历史遗忘记忆特性实现记忆均值邻域的再开采以增强算法局部开采性能。数值实验验证了不同加权学习因子函数对算法性能的差异性影响、改进算法的较好优化性能和算法稳健性等。
In order to improve global exploration and local mining performance of multivariate cosmic optimization (MVO) algorithm, an improved MVO algorithm (IMVO) is proposed to update the strategy of individuals involved in horizontal and vertical rotation. The horizontal update strategy is a horizontal migration and evolutionary mechanism based on the cosmic population level. By introducing weighted learning factors to ensure that offspring inherit location information to multiple paternal universes at the same time in order to improve individual diversity and global exploration performance, Qualitatively correct the existence of wormhole probability expression in order to ensure sufficient information exchange between individuals; longitudinal update strategy is based on the individual hierarchy of the universe a longitudinal self-learning evolution mechanism, according to the information of the optimal universe history by simulating the history of forgotten memory characteristics of cognitive Recalculation of neighborhood of memory to enhance local mining performance. Numerical experiments verify that different weighted learning factor functions have different effects on the performance of the algorithm, better optimization performance and robustness of the algorithm.