Predicting Frequency Disturbances from Wide Area Monitoring of Ambient Power System Dynamics

This paper presents a method of predicting frequency deviations caused by a disturbance (e.g. trip of a load or a generator) based on continuous monitoring of the power system. The method is based on identifying an approximate model of the dynamics of the system from ambient PMU measurements using black box system identification. Using this model, possible events of significant load variations can be simulated in order to predict the ensuing frequency disturbances. The model can be updated continuously based on ambient measurement data. The results characterize the inertial response of the system, the frequency nadir, time taken to reach the nadir, etc. Preliminary results have been demonstrated on PMU measurement data from a real power system as a part of the MIGRATE project.