论文部分内容阅读
提出一种动态环境下基于预测机制的多种群进化算法,将预测机制引入到动态进化算法的研究中,对算法所得的某些信息进行记忆,根据记忆序列构建预测模型,当环境发生变化时能够通过预测模型对动态环境进行预先判断.算法采用自组织侦查的多种群策略,多个子种群对搜索子空间进行局部搜索,主种群用于确定新的搜索子空间.在子种群的自适应调整、子种群间的拥挤操作等方面进行了改进,根据子种群所跟踪的最优解位置信息构建预测模型,当环境发生变化时通过预测及子种群的进化实现对动态环境的自适应跟踪.以移动峰问题为测试对象,实验结果表明新算法具有良好的处理动态问题的能力.
A multi-population evolutionary algorithm based on prediction mechanism in dynamic environment is proposed. The prediction mechanism is introduced into the research of dynamic evolutionary algorithm. Some information obtained by the algorithm is memorized. The prediction model is constructed according to the memory sequence. When the environment changes, The predictive model is used to predict the dynamic environment, the algorithm adopts the multi-population strategy of self-organizing detection, the multiple sub-populations search the search subspace locally, and the main population is used to determine the new search subspace.In the adaptive adjustment of sub-population, The crowding operation among subpopulations and so on is improved, the prediction model is constructed based on the information of the optimal solution locations tracked by the sub-population, and when the environment changes, adaptive tracking of the dynamic environment is realized through the prediction and the evolution of the sub-population. The peak problem is the test object, the experimental results show that the new algorithm has a good ability to deal with dynamic problems.