K-means算法的改进及其在云任务分配策略中的应用研究

发布时间：2018-06-24 22:33

本文选题：云计算 + K-means聚类　；参考：《内蒙古农业大学》2014年硕士论文

【摘要】：本文通过深入研究K-means算法,针对传统K-means算法的不足,提出将细菌觅食算法和粒子群优化算法应用于K-means算法的改进,在MATLAB环境下仿真实验,验证了改进后的K-means算法的聚类质量有明显的提高。然后通过对基于传统K-means的任务调度算法的理论研究发现,该算法所表现的调度能力有待提高。因此利用改进后的K-means算法合理分组处理任务,然后进行Min-Min算法调度任务,以达到或接近最优解即任务分配方案。通过CloudSim云平台的仿真实验,,实验结果表明,该算法不仅有效降低了任务调度的总体完成时间,而且提高了任务分配效率和系统资源的利用率,在任务完成时间和负载平衡性上优于Min-Min算法和基于传统K-means的Min-Min算法。本文的研究思路将为以后的云计算任务分配策略的研究提供参考和帮助。
[Abstract]:In this paper, through in-depth study of K-means algorithm, aiming at the shortcomings of traditional K-means algorithm, this paper proposes to apply bacterial foraging algorithm and particle swarm optimization algorithm to the improvement of K-means algorithm, and simulates the experiment in MATLAB environment. The clustering quality of the improved K-means algorithm is obviously improved. Then, through the theoretical research of the traditional K-means based task scheduling algorithm, it is found that the scheduling ability of the algorithm needs to be improved. Therefore, the improved K-means algorithm is used to deal with tasks reasonably, and then Min-Min algorithm is used to schedule tasks in order to achieve or approach the optimal solution, that is, the task allocation scheme. The simulation results of CloudSim cloud platform show that the algorithm not only reduces the overall completion time of task scheduling, but also improves the efficiency of task allocation and the utilization of system resources. It is superior to Min-Min algorithm and Min-Min algorithm based on traditional K-means in task completion time and load balance. The research ideas of this paper will provide reference and help for the future research of cloud computing task allocation strategy.
【学位授予单位】：内蒙古农业大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.09

【参考文献】