论文部分内容阅读
电力通信网故障信息的频繁项集挖掘对于电力通信网故障诊断及排除具有重要意义。依据组合学基本原理和集合理论,提出一种将频繁项集挖掘转化成统计问题的算法。该算法利用Hadoop平台,只进行1次数据库扫描,即可完成频繁项集的挖掘;当改变支持度参数时无须重新挖掘。实验结果表明:该算法具有极好的扩展性;且对于大量事务数据,算法时间开销小,对于电力通信网故障诊断具有重要意义。
The frequent itemsets mining of fault information in power communication network is of great significance to the fault diagnosis and elimination of power communication network. Based on the basic principles of combinatorics and set theory, an algorithm is proposed to convert frequent itemsets into statistical problems. The algorithm uses the Hadoop platform to perform frequent itemsets mining only once with a database scan. There is no need to re-mining the parameters when the support parameters are changed. Experimental results show that the proposed algorithm has excellent expansibility. And for a large number of transaction data, the time cost of the algorithm is small, which is of great significance for fault diagnosis of power communication network.