Confidence interval estimation for electromechanical mode parameters obtained from stochastic subspace identification

Abstract

Electromechanical modes estimated from ambient measurement signals are always subjected to statistical uncertainties. For evaluating the quality of the obtained results, it is necessary to know the respective confidence intervals of the mode estimates. A perturbation method for computing the variances of mode parameters (frequencies, damping ratios) obtained from covariance‐driven stochastic subspace identification of multiple synchrophasors is presented in this paper. The variance estimates are achieved by using the first‐order sensitivity of the mode parameter estimates to perturbations of the system matrixes, whose covariances are derived from the measured signals. The confidence interval estimation method is validated by Monte Carlo simulations with the two‐area system and the New England system. The comparison results between the mean values of the estimated variances and empirical sample variances of the estimated mode parameters show that it is feasible to estimate the confidence intervals of the mode parameters using a single set of multiple signals. The proposed method is further validated with field‐measured signals from the Western Electricity Coordinating Council system. © 2016 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc.