云环境下大规模配电网分布式最优潮流计算研究

发布时间：2019-06-14 12:19

【摘要】：智能电网通过布置大量传感器以及数据采集装置,提高对电网相关数据的采集及实时监控的能力,以实现智能化的输电和配电,是未来电网发展的必然趋势。然而,智能电网的运行却带来了所采集的数据量的爆炸性增长,具备了大数据特征。当前大规模电网的最优潮流计算中,传统的计算方法面对具备大数据特征的电力系统数据时,便出现计算速度慢,任务执行效率低等缺点,难以满足智能电网的实时计算需求;而现有的并行计算方法大多运行于专用并行机,性价比较低。因此,如何以高性价比,快速地实现最优潮流计算,成为智能电网发展中所需要解决的一个重要问题。本文研究了云环境下大规模配电网最优潮流的分布式并行计算方法。提出的方法借助Map-Reduce分布式并行编程框架,能够运行于性价比较高的Hadoop集群之上。具体来说,本文首先提出了面向Map-Reduce框架的最优潮流算法性能模型。该模型能够分析和量化在不同的集群配置下算法的执行时间,并为算法中电网的分解和计算粒度提供指导。基于此性能模型,本文提出了最优潮流计算的负载均衡算法。在给定的集群资源情况下,通过模拟退火算法确定最优的算法分解方式和计算粒度;并通过馈线重组算法实现负载均衡,从而优化最优潮流在云环境下的计算速度和效率。实验方面,本文将提出的方法与传统的串行最优潮流计算进行了比较。实验结果表明,提出的方法相对于串行方法,能够减少68.3%的计算时间。同时,本文也验证了负载均衡和不均衡情况下最优潮流算法的计算时间。实验数据表明,相比较于负载不平衡方法,本文提出的负载平衡算法能够将最优潮流的计算时间减少43.7%。
[Abstract]:By arranging a large number of sensors and data acquisition devices, smart grid can improve the ability of collecting and monitoring the related data of power grid in real time, so as to realize intelligent transmission and distribution, which is the inevitable trend of power grid development in the future. However, the operation of smart grid has brought about the explosive growth of the amount of data collected, with the characteristics of big data. At present, in the optimal power flow calculation of large-scale power grid, when the traditional calculation method faces the power system data with big data characteristics, the calculation speed is slow and the task execution efficiency is low, so it is difficult to meet the real-time computing requirements of smart grid, while most of the existing parallel computing methods run on special parallel computers, and the performance and price are relatively low. Therefore, how to realize the optimal power flow calculation quickly with high performance-price ratio has become an important problem to be solved in the development of smart grid. In this paper, the distributed parallel computing method for optimal power flow of large-scale distribution network in cloud environment is studied. With the help of Map-Reduce distributed parallel programming framework, the proposed method can run on Hadoop clusters with high performance and price. Specifically, this paper first proposes an optimal power flow algorithm performance model for Map-Reduce framework. The model can analyze and quantify the execution time of the algorithm under different cluster configurations, and provide guidance for the decomposition and calculation granularity of the power grid in the algorithm. Based on this performance model, a load balancing algorithm for optimal power flow calculation is proposed in this paper. In the case of given cluster resources, the optimal algorithm decomposition method and computational granularity are determined by simulated annealing algorithm, and the load balancing is realized by feeder reorganization algorithm, so as to optimize the computing speed and efficiency of optimal power flow in cloud environment. In the aspect of experiment, the proposed method is compared with the traditional serial optimal power flow calculation. The experimental results show that the proposed method can reduce the computing time by 68.3% compared with the serial method. At the same time, the calculation time of the optimal power flow algorithm under load balancing and imbalance is verified. The experimental data show that compared with the load imbalance method, the load balancing algorithm proposed in this paper can reduce the calculation time of the optimal power flow by 43.7%.
【学位授予单位】：华北电力大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TM744

【参考文献】