Study On The Power System Out-of-step Islanding Based On PMU

With the deepening of interconnection level and the expansion of power system, transient instability caused by disturbances may lead to large-scale blackouts, resulting in severe social and economic losses. The appearance of PMU provides a new way to analyse the problems of transient stability, so that out-of-step islanding problem can be solved online from the perspective of whole system. When the power system becomes out-of-step after large disturbances or failures, the out-of-step area can be identified quickly and accurately by online data provided by PMU. Then the out-of-step area and the rest area must be islanded at the right time and right place, thereby the oscillations can be eliminated and the safety of power system can be ensured.Firstly, the current research status of power system out-of-step islanding problem is studied and the characteristics of a variety of traditional methods are compared, including the application of artificial intelligence algorithms. In order to improve the accuracy and computation speed, the support vector machine is improved by using leave-one-out cross validation. This improved algorithm is verified through MATLAB to make sure that it can be used to analyse PMU data online.Secondly, based on the online PMU data and combined with improved SVC, a new method of power system on-line partition is proposed. With this method, the partition can be determined by the information of power angle. Combined with islanding cutest searching approach, the islanding interface of power system can be determined quickly. This method for determining islanding interface is verified by simulation of Gansu power system.At last, based on the online PMU data and combined with improved SVR, a new method for deciding islanding time is proposed. With this method, the frequency deviation of generators can be predicted by improved SVR and the power system can be islanded timely instead of too late or too early, then the safety of power system can be ensured. This method for determining islanding time is verified by simulation of Gansu power system.