摘要
为了实现电力行业多源数据的监测诊断,应对电力系统的实时处理需求,引入Storm分布式实时计算平台对数据进行处理。在此平台上部署基于混合聚类的流数据处理模型,实现数据流的故障诊断。通过将减法聚类算法和K-means算法结合实现流式数据的故障检测。减法聚类获取较优的聚类中心,而K-means算法根据此聚类中心计算出较好的分类结果。测试集群和单机的数据处理量的结果表明在集群环境下合理设置组件并行度可以提高流计算时效性和吞吐量。
In order to achieve monitoring and diagnosis of multi-source data in power industry and meet the requirements of power system real-time processing,the Storm distributed real-time computing platform was introduced to data processing.A mixed clustering based stream data processing model on this platform was deployed to achieve fault diagnosis of data streams.By combining subtractive clustering algorithm with K-means algorithm,fault detection of streaming data was achieved,in which subtractive clustering obtained optimal cluster center,and K-means algorithm calculated out better classification results according to this cluster center.By testing the data processing capacity of clusters and single machines,experiments have shown that setting component parallelism reasonably in a cluster environment can improve the efficiency and throughput of stream computing.
作者
刘少伟
LIU Shaowei(Nanjing SAC Power Grid Automation Co.,Ltd.,Nanjing 211153,China)
出处
《电工技术》
2024年第14期132-135,共4页
Electric Engineering
作者简介
刘少伟(1991-),硕士,研究方向为大数据、人工智能。
