基于属性的网格资源动态聚类研究

Attribute-based dynamic clustering of grid resources

为提高网格资源的搜索率与利用率,提出了基于属性的网格资源动态聚类法。首先以复合λ值模糊聚类树为基础,以资源簇内聚合度和资源簇间分离度为目标,构建资源聚类优化数学模型;然后运用并行遗传算法进行优化,建立了适合资源聚类模型的并行遗传算法操作流程。针对经典遗传算法存在"早熟"的不足,设计了一种新颖的矩阵交叉方法。最后,通过仿真证明了算法的高效性。

To improve the search and utilization efficiency of grid resources, an attribute-based dynamic clustering approach of grid resources was proposed. Firstly, an optimization model of resources clustering evaluated by internal clustering degree and external detached degree was proposed, which was based on fuzzy clustering tree with multi-λ,. Secondly, parallel genetic algorithm was applied to optimize the clustering results, and corresponding genetic operations were studied. Furthermore, a novel approach matrix crossover was designed to prevent prematurity. Simulation results revealed the effectiveness of the new algorithm.