PMU-based model-free method for transient instability prediction and emergency generator-shedding control

Abstract

Using PMU measurements, this paper proposes a model-free method to predict post-fault transient instability and develop emergency generator-shedding control. First, the multi-machine system is converted to an equivalent one-machine-infinite-bus (OMIB) system based on online generator clustering, and then the stability criteria are derived to judge the transient stability using the OMIB rotor speed (OMIB-ω) trajectory. Next, a new trajectory prediction algorithm based on ensemble online sequential learning machine (E-OS-ELM) is proposed to predict the OMIB-ω trajectory with an adaptive prediction window. The post-fault transient instability status can be detected in advance on basis of the predicted ω trajectory and the derived stability criteria. Lastly, when the system is foreseen to lose stability, an analytical generator-shedding control algorithm is presented, and the relationship between the generator-shedding amount and the time delay is illustrated. Case studies on the New-England 39-bus system, the NPCC 140-bus system and a realistic province power system in China are presented to show the proposed methodology can detect the instability status early, and help the system maintain synchronism.