基于PRAM模型的集群计算机混合并行算法设计

Design of Hybrid Parallel Algorithm for Cluster Computer Based on PRAM Model

为提高集群计算机混合并行计算能力,需对集群计算机混合大数据特征样本集进行优化聚类处理,提出一种基于PRAM模型的集群计算机混合并行算法。采用分块融合匹配方法进行集群计算机混合大数据特征样本线性规划处理,提取集群计算机混合大数据特征样本的统计平均特征量,结合匹配滤波检测方法进行集群计算机混合大数据特征样本集的统计分析,构建海量集群计算机混合数据序列,采用增量式支持向量机学习分类器进行数据分类,结合PRAM模型识别方法进行聚类中心自动搜索,实现集群计算机混合并行计算,提高数据聚类和并行计算的收敛性。仿真结果表明,采用该方法进行集群计算机混合并行计算的特征聚类性较高,大大减少时间开销和内存消耗,提高大数据分类检索能力。

In order to improve the hybrid parallel computing ability of cluster computer,it is necessary to optimize the clustering processing of the large data feature set of cluster computer.A hybrid parallel algorithm of cluster computer based on PRAM model is proposed.Block fusion matching is used for linear programming of mixed large data feature samples of cluster computer.The statistical average feature values of mixed large data feature samples of cluster computer are extracted.The statistical analysis of mixed large data feature samples of cluster computer is carried out by combining matched filter detection method.The mixed data sequence of massive cluster computer is constructed and incremental support vector is adopted.Computer learning classifier classifies data,and combines PRAM model recognition method to search cluster centers automatically.It realizes hybrid parallel computing of cluster computer and improves the convergence of data clustering and parallel computing.The simulation results show that this method has high clustering ability for hybrid parallel computing of cluster computers,greatly reduces the time and memory consumption,and improves the ability of large data classification and retrieval.