Empirical Noise Estimation in Distribution Synchrophasor Measurements

Noise estimation is an inadequately explored aspect of synchrophasor measurements. Though sensor components are often tested and evaluated for accuracy by the manufacturer in a lab setting, the noise in measurements made by Phasor Measurement Units deployed in the field has not been well quantified. The real world setting can have many sources of noise that are unaccounted for in the lab setting. Therefore, an empirical study of the noise level in synchrophasor measurements from an operational network is valuable. In addition to quantifying typical noise levels in the measurement data, understanding the noise model that best matches the observations is useful. In this work we present a novel test for comparing the efficacy of an additive versus a multiplicative noise model. We use three days of high-resolution voltage measurements from two distribution PMUs (D-PMUs) deployed on an operational distribution network to validate the additive noise model and estimate the typical noise level, parameterized by the measurement variance.