【摘要】 在目前配电系统相对发输电系统较为落后的电力市场环境下,低电压治理是亟须解决的问题。合理地确定治理分区是科学、有效地进行低电压治理的重要前提。基于谱聚类算法的城市低电压分区方法,首先考虑了最低电压幅值、低电压涉及户数、电压越(下)限时间及年供电量4个指标构成样本空间。其次通过欧氏距离建立拉普拉斯(Laplace)矩阵,由Laplace矩阵的相对特征值差自动确定低电压治理的分区数目,将Laplace矩阵的前两个和前三个特征向量映射到二维和三维空间,直观地划分治理分区。然后构造评估函数,评估低电压分区的效果,采用k-means算法对特征向量进行聚类,画出分区谱系图,得到层次分明的低电压分区治理方案。该算法基于复杂网络原理,具有严谨的理论依据。以江西省某市典型数据为例验证了所提算法的有效性。更多还原

【Abstract】 The governance decision making of low voltage in urban power system, while the distribution system is relatively lagging than transmission system in the power market environment, is in dire need of solving. The partition of governance is very important premise of governance decision-making of low voltage. A partition method based on spectral clustering algorithm is proposed. Firstly, sample space is established with consideration of minimum voltage amplitude, number of low voltage customer count, voltage over-limitation time and annual power supply. Secondly, Laplace matrix is built by Euclidean distance. The relative differences of eigenvalues of similarity matrix is selected automatically to determine partition number. The first two and three eigen vectors are mapped to 2d and 3d spaces which provide intuitive guidance for partition. Then evaluation function is constructed to evaluate partition effectiveness. When partition effect is acceptable, k-means algorithm is used to cluster feature vector, obtain partition spectrum diagram and structured partition scheme. The proposed algorithm is based on complex network theory and has rigorous theoretical basis. Finally, the effectiveness of proposed algorithm is verified by simulation results of a Power Grid of Jiangxi province.更多还原