上海交通大学学报

• •    下一篇

考虑虚拟化时延的配电网保护控制微服务调度模型与优化(网络首发)

  

  1. 1. 华南理工大学电力学院;2. 南方电网电力科技股份有限公司
  • 基金资助:
    广东省重点领域研发计划(2019B111109002)资助项目

Micro-Service Scheduling Model and Optimization for Protection and Control Service of Distribution Network Considering the Virtualization Delay

  1. 1. School of Electric Power Engineering, South China University of Technology, Guangzhou 510641, China; 2. China Southern Power Grid Power Technology Co., Ltd, Guangzhou 510080, China

摘要: 当前对基于微服务及容器架构配网终端的研究,忽略了虚拟化时延的影响,限制了其在保护控制等高实时性要求场景下的应用。为此,提出一种面向微服务及容器架构配网保护控制终端的微服务调度框架与策略。首先,分析了配网保护控制终端业务的时延需求与容器虚拟化时延影响并提出了保护控制微服务调度框架。然后,建立了微服务调度模型,包括业务的微服务模型、时延模型以及评价模型,并提出了基于队列状态转换的微服务调度策略。 最后,通过算例仿真对所提调度策略的有效性进行验证并进一步对调度策略进行优化,结果表明所提调度策略能够有效避免虚拟化时延的影响,且采用先到先服务算法能够取得更好的整体性能表现。

关键词: 配电网保护与控制, 虚拟化时延, 微服务, 容器, 调度策略

Abstract: The current research on distribution network terminals based on micro-service and container architecture overlooks the impact of virtualization latency in micro-service scheduling, restricting its application in scenarios with high real-time requirements such as protection and control services. To address this limitation, a micro-service scheduling framework and strategy are proposed for protection and control terminals in distribution networks based on micro-service and container architecture. Initially, an analysis is conducted on the latency performance requirements of protection and control services in the terminal and the impact of virtualization latency, leading to the formulation of a protection and control micro-service scheduling framework. Subsequently, the protection and control micro-service scheduling model is established, encompassing micro-service models, latency models, and metrics models. A micro-service scheduling strategy based on queue state transitions is then proposed. Finally, extensive case simulations are conducted, where the effectiveness of the proposed scheduling strategy is validated and Further optimization of the scheduling strategy is performed. The result indicates that the proposed scheduling strategy can effectively avoid the impact of virtualization delay, and First-Come-First-Serve (FCFS) algorithm achieves better overall performance.

Key words: protection and control for distribution network, virtualization latency, micro-service, containers, scheduling strategy

中图分类号: