Two Decentralized Dynamic State Estimation Schemes for Multimachine Power Systems with Transmission Losses

Abstract

This article addresses the problem of estimating the state of a multimachine power system (MPS). To work with power networks with lossy transmission lines, a variation of the classical third-order MPS is proposed by considering generators’ electrical power injected into the grid as a state variable. Based on a linear decentralized estimation model (tailored for a specific purpose), the state variables of generators (load angle, relative speed, and electrical power) together with terminal voltage magnitudes are estimated in a decentralized fashion through two new robustly convergent linear Luenberger estimators, one based on load angle measurement and the other based on relative speed measurement. The new MPS estimation model includes a set of robustly quick observable states, one per machine, which allows capturing the interaction with other generators, transmission line losses, unknown disturbances, and model errors. The result is a design superior to other related estimation techniques such as the extended Kalman filter (EKF) or Sliding Modes Perturbation Observer (SMPO) in terms of (i) a conventional-like simple pole placement-based tuning, (ii) low online computational load and disturbance rejection capability, and (iii) small gain-based convergence assessment. The performance of the proposed state estimation scheme is illustrated in a 3-machine power system under different operational conditions.