Ambient modal analysis of converter-dominated power systems using the multivariate autoregressive method

The increasing penetration of converter-dominated generation changes the dynamics of modern power systems. It introduces new oscillatory phenomena that challenge traditional modal analysis approaches. Accurate estimation and identification of these oscillatory modes are crucial for assessing system stability and ensuring secure operation. This paper proposes the multivariate autoregressive (MAR) method for converter-dominated power grids to estimate ambient modal parameters. The MAR method can continuously estimate modal parameters, such as frequency and damping, of converter-driven modes in real-time. This enables grid operators to monitor system stability as conditions change and take preventive actions before oscillations become problematic. Both simulation and field data are used to evaluate the performance of the method. A model replicating the west coast of the Finnish transmission network is developed and used to generate ambient oscillations through variable load disturbances. The MAR method is applied using different analysis window lengths and signal-to-noise ratio (SNR) levels. The approach is validated using phasor measurement unit (PMU) data from Fingrid Oyj. The results show that a 130-second window containing 603 cycles is sufficient for precise and consistent estimations of simulated data. For PMU measurements, a 3-minute window with approximately 878 cycles provides accurate estimations. Furthermore, a comparative analysis with established methods (ARMAX and N4SID) demonstrates that MAR delivers precise and noise-resilient modal estimates while offering exceptional computational efficiency of 0.374 ms per estimate. These findings establish MAR as a practical tool for real-time monitoring of oscillations in converter-dominated grids.