“多站融合”背景下边缘数据中心运行优化研究
Research on Operation Optimization of Edge Data Center Under the Background of “Multi-station Integration”
作者：孟超(厦门大学 能源学院，福建 厦门 361102)；刘文亮(国网厦门供电公司，福建 厦门 361000)；杨琪(厦门大学 能源学院，福建 厦门 361102)；赵英汝(厦门大学 能源学院，福建 厦门 361102)；郭熠昀(国网厦门供电公司，福建 厦门 361000)
Author：MENG Chao(College of Energy, Xiamen Univ., Xiamen 361102, China)；LIU Wenliang(State Grid Xiamen Electric Power Supply Co., Xiamen 361000, China)；YANG Qi(College of Energy, Xiamen Univ., Xiamen 361102, China)；ZHAO Yingru(College of Energy, Xiamen Univ., Xiamen 361102, China)；GUO Yiyun(State Grid Xiamen Electric Power Supply Co., Xiamen 361000, China)
收稿日期：2019-12-31 年卷（期）页码：2020,52(4):49-55
期刊名称：工程科学与技术
Journal Name：Advanced Engineering Sciences
关键字：多站融合;数据中心;数据迁移;储能电站;混合整数非线性规划
Key words：multi-station integration;data center;data migration;energy storage power station;mixed integer nonlinear programming
基金项目：国家自然科学基金项目（51876181）；国网福建省电力有限公司科技项目（5213402000HF）
中文摘要
随着泛在电力物联网的发展，传统通信技术与计算平台已逐渐无法适应庞大的数据传输与计算规模，5G通信技术与边缘计算将成为支撑泛在电力物联网建设的中坚力量。“多站融合”模式的提出为边缘数据中心的运营提供了新的方向，多站一体化运行方法有待深入研究。针对边缘数据中心的数据迁移方法及“多站融合”模式下各功能站协同优化问题，设计了涵盖数据中心、储能电站等功能站的融合站运行架构，构建了数据中心负荷模型、储能电站数学模型，并基于混合整数非线性规划方法，建立了以最小化年总成本为目标的融合站协同优化与数据迁移调度模型。通过模型的优化求解，确定了最佳数据迁移调度策略与储能电站最优运行方案。利用案例分别对数据迁移策略与储能电站的作用进行仿真分析，通过计算，采用数据迁移策略后，典型日运行成本降低了2.38%，年总成本降低了1.36%；发挥储能电站削峰填谷作用后，典型日运行成本进一步降低了8.34%，年总成本进一步降低了4.67%。结果表明：在满足时延约束情况下，将数据负荷按一定策略进行分配，可以实现各边缘数据中心互补调度，达到降低运行成本的效果；结合分时电价政策，利用储能电站参与日常电力调度，可以进一步将峰时负荷转移至谷时，从而显著降低运行成本。
英文摘要
With the development of Electrical Internet of Things (EIoT), traditional communication technologies and computing platforms have been unable to adapt to the huge scale of data transmission and computing. 5G communication technology and the edge computing will become the backbone force supporting the construction of the Electrical Internet of Things. The “multi-station integration” mode provides a new direction for the operation of edge data center, and the operation method of multi-station integration needs to be further studied. In order to study the data migration method of edge data center and the collaborative optimization problem of each function station under the mode of “multi-station integration”, a fusion station operation architecture covering data center and energy storage power station was designed in this paper. Then, a load model of data center and a mathematical model of energy storage station were constructed. Based on the mixed integer nonlinear programming method, a collaborative optimization and data migration scheduling model of integrate station with the goal of minimizing the total annual cost was established. Through solving the optimization model, the optimal data migration scheduling strategy and the optimal operation scheme of energy storage power station were determined. The simulation analysis of data migration strategy and the role of energy storage power station were carried out by case study. The calculation results showed that, the typical daily operation cost is reduced by 2.38% and the annual total cost is reduced by 1.36% after the implementation of the data migration strategy. After the energy storage power station is used to cut peak and fill valley, the typical daily operation cost is further reduced by 8.34% and the annual total cost is further reduced by 4.67%. The analysis results showed that the operation cost could be reduced when the data load is allocated according to a certain strategy under the condition of meeting the delay constraint. Combined with the time-of-use electric rate structure, the energy storage power station participating in the power dispatching could further transfer the peak load to the valley load, thus significantly reducing the operation cost.
【关闭】
