Reduced-order harmonic estimation with probable sensor failures

A nonstationary discrete-time model is used for representation of harmonic components of nonsinusoidal signals observed in electric power systems. The measurement devices utilized are assumed to have nonzero failure probabilities that introduce stochastic parameters into the dynamic model used in estimation. A finite-time, linear, unbiased, reduced-order, minimum-variance state estimator is derived to predict the future behavior of this model. This is desired in order to annihilate the harmonic components which pollute the electric power system. Simulation results are provided to illustrate the use of this estimator and support the theoretical results.