摘要
随着电网基建工程的发展,逐渐产生了大量的电网基建数据。这些数据多是结构化数据,特征是数据量庞大。针对当前电网基建数据检索效率低下的问题,提出了基于LSTM热度预测的元数据分级检索算法。结合LSTM算法,对用户访问所生成的时间序列进行预测,并通过HCM算法将元数据聚类为热点区域与非热点区域两个区域。基于预测与分区,在不同的热度分区分别进行优先级划分与文件检索构建。实验结果表明,该算法可以有效减少查询所需要的时间与空间开销,实现对电网基建数据的高效检索。
With the development of power grid infrastructure project, a large number of power grid infrastructure data has been generated gradually.These data are mostly structured data, characterized by a large amount of data.A hierarchical metadata retrieval algorithm based on LSTM heat prediction was proposed to solve the problem of low efficiency of current power grid infrastructure data retrieval.Combined with LSTM algorithm, the time series generated by user access is predicted, and the metadata is clustered into two regions, namely hot region and non-hot region, through HCM algorithm.Based on prediction and partitioning, priority division and file retrieval are constructed in different hot partitions.The experimental results show that the algorithm can effectively reduce the time and space overhead required by the query, and realize the efficient retrieval of power grid infrastructure data.
作者
荣经国
武宏波
张苏
李彧
RONG Jinguo;WU Hongbo;ZHANG Su;LI Yu(State Grid Economic Technology Research Institute,Beijing 102209,China;School of Electrical and Electronic Engineering,North China Electric Power University,Beijing 102206,China)
出处
《电子技术（上海）》
2020年第10期1-5,共5页
Electronic Technology
基金
国网经济技术研究院有限公司自主投入科技项目(ZZKJ-2020-12)
关键词
智能检索
电网基建数据
元数据检索
长短期记忆神经网络
HCM
intelligent retrieval
power grid infrastructure data
metadata retrieval
long and short term memory neural network
HCM
