Journal of Electrical Engineering and Automation
Journal of Electrical Engineering and Automation. 2025; 4: (3) ; 10.12208/j.jeea.20250078 .
总浏览量: 73
1 南京邮电大学 江苏南京
2 国网电力科学研究院有限公司 江苏南京
*通讯作者： 彭沛,单位： 南京邮电大学 江苏南京；
分布式计算是解决大规模电力系统仿真算力需求的有效方法。然而分布式计算容易受到少数问题节点性能下降所带来的负面影响，从而导致算法的计算效率严重下降。本文提出基于编码技术的仿真平台分布式计算方法，从而规避上述负面影响。首先，对构建了大规模电力系统仿真的计算任务模型；然后，通过理论分析确定了其中最优编码率的取值范围；最后，仿真分析结果表明，应用编码技术并采用合适的编码率，可有效提高大规模电力系统仿真中分布式算法的运行效率。
Distributed computing is an effective method to provide the computing sources for large-scale electric power system simulations. However, the performance degradation of some computing nodes will negatively impact distributed computing of the entire system simulations, which leads to a serious decline in the computational efficiency of algorithms. This paper investigates the simulation platforms based on coding techniques to avoid the negative effects in the distributed computing system. First, we model the computing tasks of the large-scale electric power system simulations. Then, we theoretically analyze the range of the optimal coding rate. Finally, from the simulation results, we conclude that coding with the appropriate coding rate can effectively improve the computational efficiency of the distributed computing of the large-scale electric power system simulations.
[1] 薛禹胜, 谢东亮, 薛峰, 等. 支持信息-物理-社会系统研究的跨领域交互仿真平台[J]. 电力系统自动化, 2022, 46(10): 138-148.
[2] 杨争林, 曾丹, 冯树海, 等. 电力市场实验能力建设面临的挑战及关键技术[J]. 电力系统自动化, 2022, 46(10): 111-120.
[3] Dalcin LD, Paz RR, Kler PA, et al. Parallel distributed computing using Python [J]. Advances in Water Resources, 2011,34(9): 1124-39.
[4] 刘溪桥, 刘尧, 刘行健, 等. 基于分布式计算的电网模型拼接设计与实现[J]. 微型电脑应用, 2025, 41(2): 34-37+46.
[5] 董雪涛, 冯长有, 朱子民, 等. 新型电力系统仿真工具研究初探[J]. 电力系统自动化, 2022, 46(10): 53-63.
[6] 高毅, 王成山, 李继平. 改进十字链表的稀疏矩阵技术及其在电力系统仿真中的应用[J]. 电网技术, 2011, 35(5): 33-39.
[7] 陈荃韡, 龚成明, 赵晋泉, 等. 并行稀疏系统直接求解库SuperLU_MT在状态估计中的应用[J]. 电力系统自动化, 2017, 41(3): 83-88.
[8] Karimipour H, Dinavahi V. Accelerated parallel WLS state estimation for large-scale power systems on GPU [C]. 2013 North American Power Symposium (NAPS), Manhattan, IEEE, 2013: 1-6.
[9] 郑腾飞, 周桐庆, 蔡志平, 等. 编码计算研究综述[J]. 计算机研究与发展, 2021, 58(10): 2187-2212.
[10] Wang D, Joshi G, Wornell G. Using straggler replication to reduce latency in large-scale parallel computing [J]. ACM SIGMETRICS Performance Evaluation Review, 2015, 43(3): 7-11.
[11] Aktas MF, Peng P, Soljanin E. Straggler mitigation by delayed relaunch of tasks [J]. ACM SIGMETRICS Performance Evaluation Review, 2018, 45(3): 224-31.
[12] 王艳, 王希龄, 赖宏达, 等. 面向大规模矩阵乘法的编码计算性能研究[J]. 华东交通大学学报, 2021, 38(3): 41-51.
[13] Gardner K, Zbarsky S, Doroudi S, et al. Reducing latency via redundant requests: Exact analysis [J]. ACM SIGMETRICS Performance Evaluation Review, 2015, 43(1): 347-60.
[14] Joshi G. Synergy via redundancy: Boosting service capacity with adaptive replication [J], ACM SIGMETRICS Performance Evaluation Review, 2018, 45(3):21-8.
[15] Gardner K, Harchol-Balter M, Scheller-Wolf A, et al. A better model for job redundancy: Decoupling server slowdown and job size [J]. IEEE/ACM transactions on networking, 2017, 25(6): 3353-67.
[16] Peng P, Soljanin E, Whiting P. Diversity/parallelism trade-off in distributed systems with redundancy [J], IEEE Transactions on Information Theory, 2021, 68(2): 1279-95.
[17] 张基, 谢在鹏, 毛莺池, 等. MapReduce框架下结合分布式编码计算的容错算法[J]. 计算机工程, 2021, 47(4): 173-179.
[18] 刘怡, 张磊. 基于LT码的分布式矩阵计算研究[J]. 计算机工程, 2024, 50(8): 328-335.