Fast and Accurate Real-Time Frequency Estimation Using Windowed Maximum Likelihood For Power Systems Applications

The frequency is a key indicator of the power grid’s stability and performance. Therefore, this paper introduces a window-based maximum likelihood approach for real-time frequency estimation. The proposed method consists of performing maximum likelihood frequency estimation on adjacent time windows of stream measurements to provide real-time estimation. In addition, the method is evaluated against the time window length and windowing pattern to identify the best approach based on convergence time and accuracy. Each window length and windowing pattern are tested in response to frequency step and ramp variations. Results show that maximum-likelihood estimation provides good performance under short-time observation windows and off-nominal conditions, in contrast with Fourier-based estimation techniques. Moreover, with a short time window, even simple windowing patterns yield positive outcomes.