基于KOHONEN神经网络的电压控制分区Network Partitioning Voltage Control Based on KOHONEN Neural Netwrok
刘小波,李亚玲,赵景涛,吴桢,张晓芳
摘要(Abstract)：
介绍了类别识别能力较强的KOHONEN自组织网络,并对其在二级电压控制分区方面的应用作了一些讨论和研究,将电力系统分区问题转化为几何空间中点的聚类问题。首先构造无功源坐标空间,进而取每台发电机对待分节点的灵敏程度作为样本特征量,最后引入KOHONEN神经网络进行系统分区。以新英格兰系统为例进行的数字仿真发现, KOHONEN网络是一种学习速度快、分类精度高的神经网络模型,并且适用于电压控制分区。
This paper introduces KOHONEN self-organization neural network, and gives disscussion and research in network partitioning for voltage control, thus the problem of system decomposition is converted to that of points clustering in geometric space. Firstly, a concept of Mvar coordinate space is presented, then take the sensitivity between generators and nodes as eigenvectors, at last, KOHONEN neural network is introduced to decomposition.Take the New England System as the example, the simulation results show that KOHONEN neural network is a methed which is rapid computation and strong type recognition, and it is fit for network partitioning.
关键词(KeyWords)：
无功源坐标空间; KOHONEN网络;电压控制分区;节点
Mvar coordinate space; KOHONEN neural network; voltage control area; node
基金项目(Foundation):
作者(Author):
刘小波,李亚玲,赵景涛,吴桢,张晓芳
DOI: 10.19585/j.zjdl.2007.03.001
参考文献(References)：
- [1]孙元章,王志芳,姚小寅.电力系统二级电压控制的研究[J].电力系统自动化,1999,23(9):9-14.
- [2]Lagonotte P,Sabonnadiere J C,Leost T,Y,etal.Structural Analysis of the Elestrical System:Applicationto Secondary Voltage Control in France[J].IEEE Trans onPower Systems,1989,4(2):479-486.
- [3]王耀瑜,张伯明,孙宏斌,等.一种基于专家知识的电力系统电压/无功分级分布式优化控制分区方法[J].中国电机工程学报,1998,18(3):221-224.
- [4]孙宏斌,张伯明,相年德.准稳态灵敏度的分析方法[J].中国电机工程学报,1999,19(4):9-13.
- [5]Kohonen T.Self_organization and associative memory[J].Berlin:Springer~verlag.1986.
- [6]Yang Hai,Qiao Fengxiang.NEURAL NETWORK APPOACH TO CLASSIFICATION OF TRAFFIC FLOWSTATES[J].Journal of Transportation Engineering,1998(2).
- [7]郭付军,林军.基于KOHONE网络的同杆双回线故障模式识别方法[J].继电器,2002,30(4):4-7.
- [8]Yusof S B,Rogers G J,Alden R T H.Slow CoherencyBased Network Partitioning Including Load Buses[J].IEEETrans on Power systems,1993,8(3):1375-1382.