采用受扰轨迹和独立分量分析技术识别同调机群的方法

Method of Identifying Coherent Generator Groups by Independent Component Analysis Through Perturbed Trajectories

针对大区电网互联运行中出现振荡失步时,需要正确划分同调机群以解列电网的问题,提出采用独立分量分析技术(independent component analysis,ICA)对故障后的发电机受扰轨迹进行特征提取,来识别系统中同调机群的方法。与现有方法相比,这种方法不需要获得系统元件模型和参数,而是直接球化广域测量系统(wide area measurement systems,WAMS)提供的发电机电角速度信号,再利用最大负熵准则进行独立分量分析得到特征矩阵,将高维数的多机受扰轨迹数据变换到低维空间,通过模式识别得到分群结果。8机36节点系统和西北750kV电网规划系统算例表明,该方法能有效消除噪音和坏数据的影响,准确识别出同调机群。

Identifying coherent generator groups is necessary for controlled islanding when large interconnected power systems fall into out-of-step oscillations. An independent component analysis （ICA） based feature extraction method to analyze perturbed trajectories of generators was presented. Comparing with existing solutions, new method can utilize generators＇ speed signal measured by wide area measurement systems （WAMS） directly and detailed model and parameter of power system components were not needed. After sphering measurement-data, ICA with maximal negentropy criterion was carded out to obtain feature matrix. By this means high-dimension data of multi-machine perturbed trajectories were reduced to lower dimensions, and coherent generator groups were identified by pattern recognition consequently. Simulation results of 36-bus system with 8-generator and planning northwest China 750 kV power grid show that noise signals and bad data can be eliminated. This method is accurate in identifying coherent generator groups.