基于聚类分析方法的电力系统负荷节点分区策略

摘要

In order to solve the drawback that traditional algorithms cannot indicate the crucial load buses in the specific area, a novel method is proposed to partition a group of load buses and to indicate the crucial load buses based on local voltage stability index. According to the physical significance of local voltage stability index, this method initially maps the load buses of a specific power system into the multivariate load space where the space distance is defined as the electrical distance between load buses. In the following stage, the load subgraph model is established as the equivalence of load space, then the spectrum clustering algorithm is utilized to indicate the reasonable partition for a group of load buses recursively. Subsequently, the given partition is supposed to be synthesized with local voltage stability sensitivity matrix and consequently obtain the crucial load bus in each zone. The IEEE 14-bus and IEEE 118-bus test cases are used to numerically validate the feasibility and effectiveness of the proposed method.%　　针对传统分区方法无法给出区域内关键负荷节点的缺点，基于局部电压稳定指标提出一种新的负荷节点分区及关键负荷节点确定的新方法。依据局部电压稳定指标灵敏度的物理意义，将系统中负荷节点映射到多维负荷空间中，将多维负荷空间中负荷节点间的空间距离定义为负荷节点间的电气距离，建立起与负荷节点空间相对应负荷子图模型。引入谱聚类算法，对负荷节点进行分区，层次递推的给出负荷节点的合理分区，将分区结果与局部电压稳定指标灵敏度矩阵相结合，确定各分区中的关键负荷节点。将该方法应用到 IEEE-14节点系统与 IEEE-118节点系统进行分析比较，仿真结果验证了所提算法的可行性和有效性。