Ambient-based Oscillation Mode Analysis via Dynamic Ensemble ITD and ARMA Model for Converter-based FFR Application

The accuracy of oscillation mode information become an essential reference to large-scale dynamics power system with a high proportion integration of converters. However, the difference in the data trend of the measurement unit will affect the recognition accuracy of the oscillation. To address this problem, this paper first proposes a Dynamic Ensemble Intrinsic Time-Scale Decomposition (DEITD) to remove the data trend. It improves the fitting effect of trends by optimizing the modal aliasing and number of decomposition. Next, the Yule-Walker based Auto Regressive Moving Average (ARMA) technique is utilized to estimate the low frequency oscillation modes. Multiple experiments on simulation and actual signals manifest that the proposed framework has better performance and is more real-time than some conventional methods, which can be used as the control signal of the converter-based fast frequency reserve to enhance the system stability.