关键词: 混杂多变数据, 多群落, 协作网演化, 双向驱动, 异步并行搜索

Abstract: Aiming at the hybrid variable characteristic of complex optimization problem,a multi-community bidirectional driven collaborative search method was proposed to optimize population fitness adaptively according to environmental changes.By analyzing the community characteristics of particles,the evolution model of multi-community cooperation network was established on the basis of undirected weighted graph,and the community types were divided according to the advantages and disadvantage of community fitness,in which the population node strength was evaluated by the cooperation weights and the node response between different communities.Faced with the environment changes of complex optimization problem,the adaptive performance of traditional swarm intelligence algorithm was improved by guiding the evolution of entire collaboration network from the largest node strength community.A multi-community bidirectional drive mode was constructed and its asynchronous parallel search algorithm was given,which realized the parallel evolution within community and among community and reduce the huge computation cost in the analysis of data.The experimental results showed that the presented method could rapidly adapt environmental changes of the hybrid data by optimizing the population fitness,and obtain the exact solution in an acceptable time.It provided useful means for solving the complex optimization problem.

Key words: hybrid data, multi-community, cooperation network evolution, bi-directional drive, asynchronous parallel search