关键词: 电力系统, 暂态稳定, 主动解列, 同步相量测量单元, 支持向量机, 留一交叉验证法, 同调机群, 在线识别

Abstract: When transient stability problems occur in the modern power systems, it is necessary to identify coherent generators quickly and accurately, which can provide the basis for the controlled islanding strategy in the next step. On account of dynamic trajectory information of generators measured by PMU tends to be high dimensional and nonlinear, a new method for online identification of coherent generators is proposed. Firstly, the dynamic rotor angle trajectory information can be measured by PMU. Then the information measured by PMU should be normalized to generate the standardized and high-dimensional data. After that, the optimal value of Gauss radial basis kernel function parameter g and penalty coefficient C can be determined by using LOO-CV, and the accurate classification model can be generated. Finally, this accurate classification model can be used to classify the unknown samples and get the final classification results. The simulation results show that the new method can overcome the shortcomings of the traditional methods and identify the coherent generators online both quickly and accurately, meeting the requirements of on-line analysis and real-time computing of transient stability in modern power system.

Key words: power system, transient stability, active islanding, phasor measurement unit (PMU), support vector machine (SVM), leave-one- out cross validation (LOO-CV), coherent generators, on-line identification