基于奇异值分解的同调机群识别方法

A Coherent Generators Identification Method Based on Singular Value Decomposition

提出一种基于奇异值分解的同调机群识别方法。该方法直接面向发电机实时功角数据,利用奇异值分解技术提取反映机组同调性的关键信息。从能量的角度出发,根据定义的能量贡献率指标自适应地选择奇异值数量,构造低维权矩阵,进而实现数据的显著降维。最终通过对权矩阵进行聚类分析实现同调机群识别。该方法具有原理简单、易于实现、计算量小等特点,能够十分方便地实现大型复杂电网中机群的同调识别。提出了该方法的在线应用框架,测试结果表明该方法分群速度快,具备在线应用潜力。此外,同调识别的结果可直观地进行图形展示,有利于电网的运行控制与分析。通过IEEE 39节点系统和南方电网算例验证了该方法的有效性和正确性。

This paper presented a novel approach to identify coherent generators using singular value decomposition （SVD） in power systems. The key information reflecting coherency was extracted by SVD directly from the real-time power angle values provided by wide-area measurement system （WAMS）. From the point of energy, the weight matrix with low dimensions was constructed by referring to several main singular values, the number of which was adaptively determined by energy contribution rate defined in the paper. Coherent generators were identified by applying cluster analysis on the weight matrix. The proposed method is simple, and entails small amount of calculation, so that it is suitable for complex interconnected power grid especially. Meanwhile, this paper discussed the on-line application of the method. The results show that it has high computational speed and potential to be applied online. Furthermore, the results of coherency identification can be expressed by graphs, which is beneficial to control and analysis of power systems. The effectiveness and correctness of the proposed method was validated by simulation results on IEEE39-bus system and China southern power grid （CSG）.