论文部分内容阅读
鉴于电力监控系统的多异类信源和动态融合特性,对系统海量实时监控信息进行故障规则挖掘和预测性监测。针对传统决策树法效率难提升问题,提出一种基于并行框架Map/Reduce和包含度测量相融合的规则挖掘算法(MRDT)。通过构建云计算Hadoop平台,在其分布式并行计算框架Map/Reduce基础上实现基于包含度的决策树规则挖掘算法的并行处理,高效地提取信任度较高的故障规则。以某水电站实时监控系统的电气信息为例,对MRDT算法进行实验测试,结果表明:MRDT算法在保证传统DT算法规则信任度较高的同时,提高了挖掘效率。
In view of the heterogeneous sources and dynamic convergence characteristics of power monitoring system, fault rule mining and predictive monitoring of massive real-time monitoring information of the system are conducted. Aiming at the problem of difficult to improve the efficiency of traditional decision trees, a rule mining algorithm (MRDT) based on parallel framework Map / Reduce and inclusion measure is proposed. By constructing cloud computing Hadoop platform, based on its distributed parallel computing framework Map / Reduce, parallel processing of decision tree mining algorithm based on inclusiveness is realized, and the fault rules with high trust are extracted efficiently. Taking the electrical information of a real-time monitoring system of a hydropower station as an example, the MRDT algorithm is tested experimentally. The results show that MRDT algorithm can improve the mining efficiency while ensuring the trust of the traditional DT algorithm.