基于数据流聚类的多任务并行数据控制方法  

Multitasking data control method based on data stream clustering

作　　者：张嘉慧 陈智明 黄科 王晓琪 李子龙 ZHANG Jia-hui;CHEN Zhi-ming;HUANG Ke;WANG Xiao-qi;LI Zi-long(Meizhou Power Supply Bureau of Guangdong Power Grid Co.,Ltd.,Meizhou 514021,Guangdong Province,China)

机构地区：[1]广东电网有限责任公司梅州供电局,广东梅州514021

出　　处：《信息技术》2024年第3期128-133,共6页Information Technology

基　　金：国家级新一代人工智能科技项目(2020AAA0103400)。

摘　　要：多任务并行数据的类别较多、数据流庞大,导致其控制能力较差,增加了物联网访问负担和通信开销。为了提升数据流稳定性,规范数据流访问过程,提出基于数据流聚类算法的多任务并行数据控制方法。首先利用数据采集与监视系统,从识别、采集、验证三个方面获取物联网数据流;然后计算各数据流动态趋势间的欧氏距离,使数据流相似度衡量标准统一;将优化后数据流代入聚类算法中,根据数据流时间戳合理对应任务请求;最后利用PLC控制器实现多任务并行数据控制。实验结果表明,所提方法的可扩展性高、加速比高。Various types of multitasking data and huge data stream have led to poor control ability and increase the access burden and communication overhead of the Internet of Things.In order to improve data stream stability and standardize data stream access process,a multitasking data control method based on data stream clustering algorithm is proposed.Firstly,the data collection and monitoring system is used to obtain the data stream of the Internet of Things from the three aspects of identification,collection and verification.Then,the Euclidean distance between the dynamic trends of each data stream is calculated to make the measurement standard of data stream similarity uniform.The optimized data stream is substituted into the clustering algorithm,and reasonably correspond to the task request according to the timestamp of the data stream.Finally,the PLC controller is used to realize multitasking data control.The experiment results show that the proposed method has high scalability and high speedup ratio.

关 键 词：数据采集与监视系统 数据流预处理 相似度衡量标准 聚类算法 PLC控制器 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]