A bilinear robust state estimator

This paper proposes a bilinear robust state estimator (BRSE), which includes a linear programming problem and a quadratic programming problem with a nonlinear transformation in between. The main advantages of BRSE are threefold: (i) it can automatically suppress bad measurements in the estimation process effectively, thus possessing good robustness; (ii) it is globally convex, theoretically guaranteeing a global optimization for state estimator; and (iii) because no nonlinear iterative algorithms are required to solve the linearized model, there is no convergence problem. To further improve BRSE's ability to inhibit leverage bad measurements, this paper also proposes a robust weighted least absolute value estimation with optimal transformations, which can be directly used in BRSE. Simulations are conducted on a rudimentary 3-bus system for validating the proposed methodology and algorithm. Furthermore, benchmark systems including IEEE 9, 14, 30, 39, 57, 118, and 300-bus systems are tested to demonstrate the efficiency and reliability of our methodology and algorithm. Copyright (c) 2015 John Wiley & Sons, Ltd.