Ambient signals based power system oscillation modes identification considering model order selection

The estimation of oscillation modes is important to the monitoring and damping of low frequency oscillation in power system. Ambient data caused by low level stochastic disturbances can be used to identify the low frequency oscillation properties. In this paper, the autoregressive moving averaging(ARMA) method is used to analyze the ambient data. As a key step in the ARMA method, the selection of model order is primarily studied. By comparing different model order selection criterions, Bayesian Information Criterion(BIC) is chosen to determine the model order, and ARMA (2n, 2n−1) modeling procedure is adopted to improve the calculation efficiency. The overall flowchart of the proposed low frequency oscillation analysis based on ambient data is also given, which is designed for the online application. The advantages of this approach are validated through simulations in 36-node benchmark system and practical ambient signals measured in China Southern Power Grid.