基于聚类的多运行方式下电力系统稳定器设计

Power System Stabilizers Design under Multi-Operation Conditions Based on Clustering

将模式识别中的聚类算法用于电力系统多运行方式的分类,提出了一种多运行方式下电力系统稳定器的设计方法.选择所有节点注入功率或所有机电振荡阻尼比为表征系统运行方式的特征向量,基于基本顺序聚类算法确定类的数目,用层次聚类和C均值聚类算法得到每类的中心运行方式.电力系统稳定器的参数设计仅针对各类的中心运行方式进行,将此问题转化为一个非光滑优化问题,用信赖域法加以求解,求得的参数可用于该类别其余运行方式.示例系统的仿真结果表明,提出的方法可以显著降低多运行方式下控制器参数设计问题的规模,从而大大减小计算量.

The clustering algorithm in pattern recognition is applied to classify multi-operating conditions of power system, and a novel design method of power system stabilizers under multi-operating conditions is proposed. The feature vector to measure operating condition consists of the injected power of all nodes or damping ratio of all electromechanical modes. The class number is determined by the basic sequential algorithmic scheme. Hierarchical clustering and C-mean clustering algorithms are applied to obtain the central operating condition. The parameters of power system stabilizers are designed only for each central operating condition. By taking the design problem as a non-smooth optimization problem, the trust-region algorithm is adopted for solution,and the obtained parameters can be applied to the other operating conditions in the same cluster. The simulation results of test systems show that the proposed approach enables to lower the scale of parameter design under multi-operating conditions, remarkably reducing the computational tasks.