基于半监督谱聚类的最优主动解列断面搜索

发布时间：2018-03-06 08:31

本文选题：主动解列　切入点：NP难题　出处：《电网技术》2015年01期 　论文类型：期刊论文

【摘要】：主动解列最优断面搜索是依据广域测量信息,在大电网遭受大扰动失步崩溃之前,依据实时工况和运行方式,快速准确求取电力孤岛划分的紧急策略。然而,在实际大系统的求解中,计算复杂度呈几何指数增长,是一个NP难题。提出了一种半监督谱聚类算法,首先采用最小复合有功潮流冲击的目标函数和机组同调/分离等相关约束构建详细解列断面搜索模型,然后将最优断面搜索的优化求解过程,映射为约束谱聚类对静态图分割的松弛解求取过程,最后通过改进的PAM聚类算法选择最优主动解列断面。上述过程,在不丢失全网信息前提下,降低了时间复杂度,IEEE 118标准算例和四川电网实际系统的仿真验证,证明了该算法的正确性、有效性和快速性。
[Abstract]:The optimal cross-section search for active decoupling is based on wide-area measurement information, and the emergency strategy of power island division is obtained quickly and accurately according to real-time working conditions and operation mode before large power grid suffers from large disturbance and collapse. In the real large-scale system, the computational complexity increases exponentially, which is a NP problem. A semi-supervised spectral clustering algorithm is proposed. Firstly, using the objective function of the minimum complex active power flow impact and the related constraints such as unit homology / separation, the detailed dissolving section search model is constructed, and then the optimization process of the optimal section search is put forward. The mapping is the relaxed solution of static graph segmentation by constrained spectral clustering. Finally, the optimal active disassembly section is selected by the improved PAM clustering algorithm. The time complexity of IEEE 118 standard is reduced and the simulation results of Sichuan power system are given. The correctness, validity and rapidity of the proposed algorithm are proved.
【作者单位】：武汉大学电气工程学院;淮安供电公司;国电南瑞科技股份有限公司;
【基金】：中央高校基本科研业务费专项资金资助博士后科学基金(2014M552080)~~
【分类号】：TM73

【相似文献】