电力系统及其自动化学报2017，Vol.29Issue(7)：27-33,7.DOI:10.3969/j.issn.1003-8930.2017.07.005
分布式配电网监测的流计算实时并行处理技术
Real-time Parallel Processing Technique in Stream Computing for Distributed Monitoring of Distribution Network
摘要
Abstract
Aiming at the serious latency problem due to the massive data's storage before processing in the real-time monitoring of distribution network,a fast processing method for monitoring the information flow is proposed based on Storm distributed stream computing framework. With the combination of stream computing and full-distributed memory computing,a topology instance for monitoring the data flow is designed to realize cluster processing in the stream com?puting of monitoring the information from distribution network,by using a parallel programming model of topology com?ponents and stream computing of node cluster. A railway 10 kV distribution network monitoring system is taken as an ex?ample,and it is used to conduct a cluster test for the topology instance composed of different spouts and bolts,which verifies that stream computing can improve the processing efficiency of real-time monitoring by processing the data be?fore storage. Results show that under a cluster environment,the topology instance correctly outputs the processing re?sults of topology components,and the latency for monitoring data can be controlled to be within hundreds of millisec?onds by carrying out memory computing before data storage.关键词
配电网/处理延迟/分布式监测/流计算/拓扑实例Key words
distribution network/processing latency/distributed monitoring/stream computing/topology instance分类
信息技术与安全科学引用本文复制引用
屈志坚,王冬..分布式配电网监测的流计算实时并行处理技术[J].电力系统及其自动化学报,2017,29(7):27-33,7.基金项目
国家自然科学基金资助项目(51567008,51267005) （51567008,51267005）
江西省自然科学基金资助项目(20161BAB206156) （20161BAB206156）
江西省杰出青年人才资助计划项目(20162BCB23045) （20162BCB23045）
