Instrumental Variable-Based Recursive Least-Squares Method to Estimate Power System Modal Characteristics

Time-domain parametric algorithms experience estimation bias in the presence of colored noise when the AutoRegressive (AR) parametric model is used. On the other hand, the AutoRegressive Moving Average (ARMA) model requires determining both the AR and MA orders to achieve unbiased estimation, which is often difficult to tune. This paper aims to provide an online unbiased estimation of power system modal characteristics in the presence of colored measurement noise using a novel instrumental variable-based recursive least squares algorithm. To do so, the online estimate of the system's noise-free output is used as an instrumental variable in the RLS algorithm, and AR model coefficients are determined. The applicability and accuracy of the proposed method are evaluated using simulated ambient data contaminated with colored noise in the IEEE 16-machine, 5-area test system excited by Gaussian random load modulations. The results demonstrate the accuracy and effectiveness of the proposed method.