面向海量电力光电监测时序数据交互式分析的时空关联索引与渐进式近似查询引擎
DOI:
CSTR:
作者:
作者单位:

广东电网有限责任公司

作者简介:

通讯作者:

中图分类号:

基金项目:

2025年数字技术平台适应性改造(三区数据中心(省侧)能力升级)(037800HK24120111)


Spatiotemporal correlation index and progressive approximate query engine for interactive analysis of massive power opto-electronic monitoring time-series data
Author:
Affiliation:

Fund Project:

  • 摘要
  • |
  • 图/表
  • |
  • 访问统计
  • |
  • 参考文献
  • |
  • 相似文献
  • |
  • 引证文献
  • |
  • 资源附件
  • |
  • 文章评论
    摘要:

    随着分布式光伏出力监测、光电式状态传感等技术在电力系统的规模化部署,海量电力光电监测数据的高效分析成为运维关键。为平衡海量电力光电时序数据查询的延迟与精度矛盾,本文提出一种名为Proxima的渐进式近似查询引擎。其核心创新是设计了基于时空关联的双层混合索引与GPU加速的渐进式计算流水线。查询时,引擎首先在毫秒级内返回基于数据的近似结果与置信区间,随后在后台进行全量精确计算并持续优化。实验证明,在万亿级光电监测数据点规模上,Proxima能在100毫秒内返回误差率低于2%的近似结果,最终精确结果的生成速度亦较现有分布式计算框架提升一个数量级,显著提升了电力光电监测数据交互式分析的体验。

    Abstract:

    With the large-scale deployment of technologies such as distributed photovoltaic output monitoring and optoelec-tronic status sensing in power systems, the efficient analysis of massive power optoelectronic monitoring data has become crucial for operation and maintenance. To balance the contradiction between delay and accuracy in query-ing massive power optoelectronic monitoring time-series data, this paper proposes a progressive approximate query engine called Proxima. Its core innovation is... Experimental results have shown that on a scale of trillions of opto-electronic monitoring data points, Proxima can return approximate results with an error rate of less than 2% within 100 milliseconds, and the generation speed of the final accurate results is also one order of magnitude faster than existing distributed computing frameworks, significantly improving the interactive analysis experience of power optoelectronic monitoring data.

    参考文献
    相似文献
    引证文献
引用本文
分享
相关视频

文章指标
  • 点击次数:
  • 下载次数:
  • HTML阅读次数:
  • 引用次数:
历史
  • 收稿日期:2026-07-06
  • 最后修改日期:2026-07-06
  • 录用日期:2026-07-08
  • 在线发布日期:
  • 出版日期:
文章二维码